Merge in upstream headscale changes

.github/workflows/docs.yml

@@ -26,7 +26,7 @@ jobs:
           key: ${{ github.ref }}
           path: .cache
       - name: Setup dependencies
-        run: pip install mkdocs-material pillow cairosvg mkdocs-minify-plugin
+        run: pip install -r docs/requirements.txt
       - name: Build docs
         run: mkdocs build --strict
       - name: Upload artifact

.github/workflows/stale.yml

@@ -12,10 +12,10 @@ jobs:
     steps:
       - uses: actions/stale@v5
         with:
-          days-before-issue-stale: 180
+          days-before-issue-stale: 90
-          days-before-issue-close: 14
+          days-before-issue-close: 7
           stale-issue-label: "stale"
-          stale-issue-message: "This issue is stale because it has been open for 180 days with no activity."
+          stale-issue-message: "This issue is stale because it has been open for 90 days with no activity."
           close-issue-message: "This issue was closed because it has been inactive for 14 days since being marked as stale."
           days-before-pr-stale: -1
           days-before-pr-close: -1

.github/workflows/test-integration-v2-TestACLAllowStarDst.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLAllowStarDst
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestACLAllowUser80Dst.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLAllowUser80Dst
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestACLAllowUserDst.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLAllowUserDst
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestACLDenyAllPort80.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLDenyAllPort80
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestACLDevice1CanAccessDevice2.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLDevice1CanAccessDevice2
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestACLHostsInNetMapTable.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLHostsInNetMapTable
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestACLNamedHostsCanReach.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLNamedHostsCanReach
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestACLNamedHostsCanReachBySubnet.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestACLNamedHostsCanReachBySubnet
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestApiKeyCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestApiKeyCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestAuthKeyLogoutAndRelogin.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestAuthKeyLogoutAndRelogin
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestAuthWebFlowAuthenticationPingAll.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestAuthWebFlowAuthenticationPingAll
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestAuthWebFlowLogoutAndRelogin.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestAuthWebFlowLogoutAndRelogin
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestCreateTailscale.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestCreateTailscale
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestDERPServerScenario.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestDERPServerScenario
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestEnablingRoutes.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestEnablingRoutes
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestEphemeral.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestEphemeral
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestExpireNode.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestExpireNode
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestHASubnetRouterFailover.yaml

@@ -0,0 +1,67 @@
+# DO NOT EDIT, generated with cmd/gh-action-integration-generator/main.go
+# To regenerate, run "go generate" in cmd/gh-action-integration-generator/
+
+name: Integration Test v2 - TestHASubnetRouterFailover
+
+on: [pull_request]
+
+concurrency:
+  group: ${{ github.workflow }}-$${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  TestHASubnetRouterFailover:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+
+      - uses: DeterminateSystems/nix-installer-action@main
+      - uses: DeterminateSystems/magic-nix-cache-action@main
+      - uses: satackey/action-docker-layer-caching@main
+        continue-on-error: true
+
+      - name: Get changed files
+        id: changed-files
+        uses: tj-actions/changed-files@v34
+        with:
+          files: |
+            *.nix
+            go.*
+            **/*.go
+            integration_test/
+            config-example.yaml
+
+      - name: Run TestHASubnetRouterFailover
+        uses: Wandalen/wretry.action@master
+        if: steps.changed-files.outputs.any_changed == 'true'
+        with:
+          attempt_limit: 5
+          command: |
+            nix develop --command -- docker run \
+              --tty --rm \
+              --volume ~/.cache/hs-integration-go:/go \
+              --name headscale-test-suite \
+              --volume $PWD:$PWD -w $PWD/integration \
+              --volume /var/run/docker.sock:/var/run/docker.sock \
+              --volume $PWD/control_logs:/tmp/control \
+              golang:1 \
+                go run gotest.tools/gotestsum@latest -- ./... \
+                  -failfast \
+                  -timeout 120m \
+                  -parallel 1 \
+                  -run "^TestHASubnetRouterFailover$"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: logs
+          path: "control_logs/*.log"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: pprof
+          path: "control_logs/*.pprof.tar"

.github/workflows/test-integration-v2-TestHeadscale.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestHeadscale
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestNodeAdvertiseTagNoACLCommand.yaml

@@ -0,0 +1,67 @@
+# DO NOT EDIT, generated with cmd/gh-action-integration-generator/main.go
+# To regenerate, run "go generate" in cmd/gh-action-integration-generator/
+
+name: Integration Test v2 - TestNodeAdvertiseTagNoACLCommand
+
+on: [pull_request]
+
+concurrency:
+  group: ${{ github.workflow }}-$${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  TestNodeAdvertiseTagNoACLCommand:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+
+      - uses: DeterminateSystems/nix-installer-action@main
+      - uses: DeterminateSystems/magic-nix-cache-action@main
+      - uses: satackey/action-docker-layer-caching@main
+        continue-on-error: true
+
+      - name: Get changed files
+        id: changed-files
+        uses: tj-actions/changed-files@v34
+        with:
+          files: |
+            *.nix
+            go.*
+            **/*.go
+            integration_test/
+            config-example.yaml
+
+      - name: Run TestNodeAdvertiseTagNoACLCommand
+        uses: Wandalen/wretry.action@master
+        if: steps.changed-files.outputs.any_changed == 'true'
+        with:
+          attempt_limit: 5
+          command: |
+            nix develop --command -- docker run \
+              --tty --rm \
+              --volume ~/.cache/hs-integration-go:/go \
+              --name headscale-test-suite \
+              --volume $PWD:$PWD -w $PWD/integration \
+              --volume /var/run/docker.sock:/var/run/docker.sock \
+              --volume $PWD/control_logs:/tmp/control \
+              golang:1 \
+                go run gotest.tools/gotestsum@latest -- ./... \
+                  -failfast \
+                  -timeout 120m \
+                  -parallel 1 \
+                  -run "^TestNodeAdvertiseTagNoACLCommand$"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: logs
+          path: "control_logs/*.log"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: pprof
+          path: "control_logs/*.pprof.tar"

.github/workflows/test-integration-v2-TestNodeAdvertiseTagWithACLCommand.yaml

@@ -0,0 +1,67 @@
+# DO NOT EDIT, generated with cmd/gh-action-integration-generator/main.go
+# To regenerate, run "go generate" in cmd/gh-action-integration-generator/
+
+name: Integration Test v2 - TestNodeAdvertiseTagWithACLCommand
+
+on: [pull_request]
+
+concurrency:
+  group: ${{ github.workflow }}-$${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  TestNodeAdvertiseTagWithACLCommand:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+
+      - uses: DeterminateSystems/nix-installer-action@main
+      - uses: DeterminateSystems/magic-nix-cache-action@main
+      - uses: satackey/action-docker-layer-caching@main
+        continue-on-error: true
+
+      - name: Get changed files
+        id: changed-files
+        uses: tj-actions/changed-files@v34
+        with:
+          files: |
+            *.nix
+            go.*
+            **/*.go
+            integration_test/
+            config-example.yaml
+
+      - name: Run TestNodeAdvertiseTagWithACLCommand
+        uses: Wandalen/wretry.action@master
+        if: steps.changed-files.outputs.any_changed == 'true'
+        with:
+          attempt_limit: 5
+          command: |
+            nix develop --command -- docker run \
+              --tty --rm \
+              --volume ~/.cache/hs-integration-go:/go \
+              --name headscale-test-suite \
+              --volume $PWD:$PWD -w $PWD/integration \
+              --volume /var/run/docker.sock:/var/run/docker.sock \
+              --volume $PWD/control_logs:/tmp/control \
+              golang:1 \
+                go run gotest.tools/gotestsum@latest -- ./... \
+                  -failfast \
+                  -timeout 120m \
+                  -parallel 1 \
+                  -run "^TestNodeAdvertiseTagWithACLCommand$"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: logs
+          path: "control_logs/*.log"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: pprof
+          path: "control_logs/*.pprof.tar"

.github/workflows/test-integration-v2-TestNodeCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestNodeCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestNodeExpireCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestNodeExpireCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestNodeMoveCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestNodeMoveCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestNodeOnlineLastSeenStatus.yaml

@@ -0,0 +1,67 @@
+# DO NOT EDIT, generated with cmd/gh-action-integration-generator/main.go
+# To regenerate, run "go generate" in cmd/gh-action-integration-generator/
+
+name: Integration Test v2 - TestNodeOnlineLastSeenStatus
+
+on: [pull_request]
+
+concurrency:
+  group: ${{ github.workflow }}-$${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  TestNodeOnlineLastSeenStatus:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+
+      - uses: DeterminateSystems/nix-installer-action@main
+      - uses: DeterminateSystems/magic-nix-cache-action@main
+      - uses: satackey/action-docker-layer-caching@main
+        continue-on-error: true
+
+      - name: Get changed files
+        id: changed-files
+        uses: tj-actions/changed-files@v34
+        with:
+          files: |
+            *.nix
+            go.*
+            **/*.go
+            integration_test/
+            config-example.yaml
+
+      - name: Run TestNodeOnlineLastSeenStatus
+        uses: Wandalen/wretry.action@master
+        if: steps.changed-files.outputs.any_changed == 'true'
+        with:
+          attempt_limit: 5
+          command: |
+            nix develop --command -- docker run \
+              --tty --rm \
+              --volume ~/.cache/hs-integration-go:/go \
+              --name headscale-test-suite \
+              --volume $PWD:$PWD -w $PWD/integration \
+              --volume /var/run/docker.sock:/var/run/docker.sock \
+              --volume $PWD/control_logs:/tmp/control \
+              golang:1 \
+                go run gotest.tools/gotestsum@latest -- ./... \
+                  -failfast \
+                  -timeout 120m \
+                  -parallel 1 \
+                  -run "^TestNodeOnlineLastSeenStatus$"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: logs
+          path: "control_logs/*.log"
+
+      - uses: actions/upload-artifact@v3
+        if: always() && steps.changed-files.outputs.any_changed == 'true'
+        with:
+          name: pprof
+          path: "control_logs/*.pprof.tar"

.github/workflows/test-integration-v2-TestNodeRenameCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestNodeRenameCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestNodeTagCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestNodeTagCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestOIDCAuthenticationPingAll.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestOIDCAuthenticationPingAll
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestOIDCExpireNodesBasedOnTokenExpiry.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestOIDCExpireNodesBasedOnTokenExpiry
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestPingAllByHostname.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestPingAllByHostname
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestPingAllByIP.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestPingAllByIP
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestPreAuthKeyCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestPreAuthKeyCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestPreAuthKeyCommandReusableEphemeral.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestPreAuthKeyCommandReusableEphemeral
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestPreAuthKeyCommandWithoutExpiry.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestPreAuthKeyCommandWithoutExpiry
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestResolveMagicDNS.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestResolveMagicDNS
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestSSHIsBlockedInACL.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestSSHIsBlockedInACL
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestSSHMultipleUsersAllToAll.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestSSHMultipleUsersAllToAll
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestSSHNoSSHConfigured.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestSSHNoSSHConfigured
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestSSHOneUserToAll.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestSSHOneUserToAll
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestSSHUserOnlyIsolation.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestSSHUserOnlyIsolation
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestTaildrop.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestTaildrop
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestTailscaleNodesJoiningHeadcale.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestTailscaleNodesJoiningHeadcale
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.github/workflows/test-integration-v2-TestUserCommand.yaml

@@ -35,8 +35,11 @@ jobs:
             config-example.yaml
 
       - name: Run TestUserCommand
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -46,7 +49,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

.gitignore

@@ -1,5 +1,6 @@
 ignored/
 tailscale/
+.vscode/
 
 # Binaries for programs and plugins
 *.exe

CHANGELOG.md

@@ -23,11 +23,18 @@ after improving the test harness as part of adopting [#1460](https://github.com/
 
 ### BREAKING
 
-Code reorganisation, a lot of code has moved, please review the following PRs accordingly [#1473](https://github.com/juanfont/headscale/pull/1473)
+- Code reorganisation, a lot of code has moved, please review the following PRs accordingly [#1473](https://github.com/juanfont/headscale/pull/1473)
-API: Machine is now Node [#1553](https://github.com/juanfont/headscale/pull/1553)
+- API: Machine is now Node [#1553](https://github.com/juanfont/headscale/pull/1553)
+- Remove support for older Tailscale clients [#1611](https://github.com/juanfont/headscale/pull/1611)
+  - The latest supported client is 1.36
+- Headscale checks that _at least_ one DERP is defined at start [#1564](https://github.com/juanfont/headscale/pull/1564)
+  - If no DERP is configured, the server will fail to start, this can be because it cannot load the DERPMap from file or url.
+- Embedded DERP server requires a private key [#1611](https://github.com/juanfont/headscale/pull/1611)
+  - Add a filepath entry to [`derp.server.private_key_path`](https://github.com/juanfont/headscale/blob/b35993981297e18393706b2c963d6db882bba6aa/config-example.yaml#L95)
 
 ### Changes
 
+Use versioned migrations [#1644](https://github.com/juanfont/headscale/pull/1644)
 Make the OIDC callback page better [#1484](https://github.com/juanfont/headscale/pull/1484)
 Allow use of the username OIDC claim [#1287](https://github.com/juanfont/headscale/pull/1287)
 SSH support [#1487](https://github.com/juanfont/headscale/pull/1487)
@@ -36,6 +43,7 @@ Use error group handling to ensure tests actually pass [#1535](https://github.co
 Fix hang on SIGTERM [#1492](https://github.com/juanfont/headscale/pull/1492) taken from [#1480](https://github.com/juanfont/headscale/pull/1480)
 Send logs to stderr by default [#1524](https://github.com/juanfont/headscale/pull/1524)
 Fix [TS-2023-006](https://tailscale.com/security-bulletins/#ts-2023-006) security UPnP issue [#1563](https://github.com/juanfont/headscale/pull/1563)
+Turn off gRPC logging [#1640](https://github.com/juanfont/headscale/pull/1640) fixes [#1259](https://github.com/juanfont/headscale/issues/1259)
 Add `oidc.groups_claim`, `oidc.email_claim`, and `oidc.username_claim` to allow setting those claim names [#1594](https://github.com/juanfont/headscale/pull/1594)
 
 ## 0.22.3 (2023-05-12)

Dockerfile

@@ -9,7 +9,7 @@ RUN go mod download
 
 COPY . .
 
-RUN CGO_ENABLED=0 GOOS=linux go install -tags ts2019 -ldflags="-s -w -X github.com/juanfont/headscale/cmd/headscale/cli.Version=$VERSION" -a ./cmd/headscale
+RUN CGO_ENABLED=0 GOOS=linux go install -ldflags="-s -w -X github.com/juanfont/headscale/cmd/headscale/cli.Version=$VERSION" -a ./cmd/headscale
 RUN strip /go/bin/headscale
 RUN test -e /go/bin/headscale
 
@@ -17,9 +17,9 @@ RUN test -e /go/bin/headscale
 FROM docker.io/debian:bookworm-slim
 
 RUN apt-get update \
-    && apt-get install -y ca-certificates \
+  && apt-get install -y ca-certificates \
-    && rm -rf /var/lib/apt/lists/* \
+  && rm -rf /var/lib/apt/lists/* \
-    && apt-get clean
+  && apt-get clean
 
 COPY --from=build /go/bin/headscale /bin/headscale
 ENV TZ UTC

Dockerfile.debug

@@ -9,7 +9,7 @@ RUN go mod download
 
 COPY . .
 
-RUN CGO_ENABLED=0 GOOS=linux go install -tags ts2019 -ldflags="-s -w -X github.com/juanfont/headscale/cmd/headscale/cli.Version=$VERSION" -a ./cmd/headscale
+RUN CGO_ENABLED=0 GOOS=linux go install -ldflags="-s -w -X github.com/juanfont/headscale/cmd/headscale/cli.Version=$VERSION" -a ./cmd/headscale
 RUN test -e /go/bin/headscale
 
 # Debug image
@@ -19,9 +19,9 @@ COPY --from=build /go/bin/headscale /bin/headscale
 ENV TZ UTC
 
 RUN apt-get update \
-    && apt-get install --no-install-recommends --yes less jq \
+  && apt-get install --no-install-recommends --yes less jq \
-    && rm -rf /var/lib/apt/lists/* \
+  && rm -rf /var/lib/apt/lists/* \
-    && apt-get clean
+  && apt-get clean
 RUN mkdir -p /var/run/headscale
 
 # Need to reset the entrypoint or everything will run as a busybox script

Makefile

@@ -10,8 +10,6 @@ ifeq ($(filter $(GOOS), openbsd netbsd soloaris plan9), )
 else
 endif
 
-TAGS = -tags ts2019
-
 # GO_SOURCES = $(wildcard *.go)
 # PROTO_SOURCES = $(wildcard **/*.proto)
 GO_SOURCES = $(call rwildcard,,*.go)
@@ -24,7 +22,7 @@ build:
 dev: lint test build
 
 test:
-	gotestsum -- $(TAGS) -short -coverprofile=coverage.out ./...
+	gotestsum -- -short -coverprofile=coverage.out ./...
 
 test_integration:
 	docker run \
@@ -34,7 +32,7 @@ test_integration:
 		-v $$PWD:$$PWD -w $$PWD/integration \
 		-v /var/run/docker.sock:/var/run/docker.sock \
 		golang:1 \
-		go run gotest.tools/gotestsum@latest -- $(TAGS) -failfast ./... -timeout 120m -parallel 8
+		go run gotest.tools/gotestsum@latest -- -failfast ./... -timeout 120m -parallel 8
 
 lint:
 	golangci-lint run --fix --timeout 10m

README.md

@@ -466,6 +466,13 @@ make build
             <sub style="font-size:14px"><b>unreality</b></sub>
         </a>
     </td>
+    <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
+        <a href=https://github.com/vsychov>
+            <img src=https://avatars.githubusercontent.com/u/2186303?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=MichaelKo/>
+            <br />
+            <sub style="font-size:14px"><b>MichaelKo</b></sub>
+        </a>
+    </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/kevin1sMe>
             <img src=https://avatars.githubusercontent.com/u/6886076?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=kevinlin/>
@@ -473,6 +480,8 @@ make build
             <sub style="font-size:14px"><b>kevinlin</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/QZAiXH>
             <img src=https://avatars.githubusercontent.com/u/23068780?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Snack/>
@@ -480,8 +489,6 @@ make build
             <sub style="font-size:14px"><b>Snack</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/artemklevtsov>
             <img src=https://avatars.githubusercontent.com/u/603798?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Artem Klevtsov/>
@@ -517,6 +524,8 @@ make build
             <sub style="font-size:14px"><b>LIU HANCHENG</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/motiejus>
             <img src=https://avatars.githubusercontent.com/u/107720?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Motiejus Jakštys/>
@@ -524,8 +533,6 @@ make build
             <sub style="font-size:14px"><b>Motiejus Jakštys</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/pvinis>
             <img src=https://avatars.githubusercontent.com/u/100233?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Pavlos Vinieratos/>
@@ -547,13 +554,6 @@ make build
             <sub style="font-size:14px"><b>Steven Honson</b></sub>
         </a>
     </td>
-    <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
-        <a href=https://github.com/vsychov>
-            <img src=https://avatars.githubusercontent.com/u/2186303?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=MichaelKo/>
-            <br />
-            <sub style="font-size:14px"><b>MichaelKo</b></sub>
-        </a>
-    </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/ratsclub>
             <img src=https://avatars.githubusercontent.com/u/25647735?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Victor Freire/>
@@ -577,6 +577,13 @@ make build
             <sub style="font-size:14px"><b>thomas</b></sub>
         </a>
     </td>
+    <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
+        <a href=https://github.com/puzpuzpuz>
+            <img src=https://avatars.githubusercontent.com/u/37772591?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Andrei Pechkurov/>
+            <br />
+            <sub style="font-size:14px"><b>Andrei Pechkurov</b></sub>
+        </a>
+    </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/linsomniac>
             <img src=https://avatars.githubusercontent.com/u/466380?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Sean Reifschneider/>
@@ -598,13 +605,6 @@ make build
             <sub style="font-size:14px"><b>Albert Copeland</b></sub>
         </a>
     </td>
-    <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
-        <a href=https://github.com/puzpuzpuz>
-            <img src=https://avatars.githubusercontent.com/u/37772591?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Andrei Pechkurov/>
-            <br />
-            <sub style="font-size:14px"><b>Andrei Pechkurov</b></sub>
-        </a>
-    </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/theryecatcher>
             <img src=https://avatars.githubusercontent.com/u/16442416?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Anoop Sundaresh/>
@@ -658,6 +658,13 @@ make build
     </td>
 </tr>
 <tr>
+    <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
+        <a href=https://github.com/winterheart>
+            <img src=https://avatars.githubusercontent.com/u/81112?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Azamat H. Hackimov/>
+            <br />
+            <sub style="font-size:14px"><b>Azamat H. Hackimov</b></sub>
+        </a>
+    </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/stensonb>
             <img src=https://avatars.githubusercontent.com/u/933389?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Bryan Stenson/>
@@ -693,6 +700,8 @@ make build
             <sub style="font-size:14px"><b>Felix Kronlage-Dammers</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/felixonmars>
             <img src=https://avatars.githubusercontent.com/u/1006477?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Felix Yan/>
@@ -700,8 +709,6 @@ make build
             <sub style="font-size:14px"><b>Felix Yan</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/gabe565>
             <img src=https://avatars.githubusercontent.com/u/7717888?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Gabe Cook/>
@@ -723,6 +730,13 @@ make build
             <sub style="font-size:14px"><b>hrtkpf</b></sub>
         </a>
     </td>
+    <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
+        <a href=https://github.com/jessebot>
+            <img src=https://avatars.githubusercontent.com/u/2389292?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=JesseBot/>
+            <br />
+            <sub style="font-size:14px"><b>JesseBot</b></sub>
+        </a>
+    </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/jimt>
             <img src=https://avatars.githubusercontent.com/u/180326?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Jim Tittsler/>
@@ -730,6 +744,8 @@ make build
             <sub style="font-size:14px"><b>Jim Tittsler</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/jsiebens>
             <img src=https://avatars.githubusercontent.com/u/499769?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Johan Siebens/>
@@ -744,8 +760,6 @@ make build
             <sub style="font-size:14px"><b>John Axel Eriksson</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/ShadowJonathan>
             <img src=https://avatars.githubusercontent.com/u/22740616?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Jonathan de Jong/>
@@ -774,6 +788,8 @@ make build
             <sub style="font-size:14px"><b>Lucalux</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/foxtrot>
             <img src=https://avatars.githubusercontent.com/u/4153572?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Marc/>
@@ -788,8 +804,6 @@ make build
             <sub style="font-size:14px"><b>Mesar Hameed</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/mikejsavage>
             <img src=https://avatars.githubusercontent.com/u/579299?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Michael Savage/>
@@ -818,6 +832,8 @@ make build
             <sub style="font-size:14px"><b>Pontus N</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/nnsee>
             <img src=https://avatars.githubusercontent.com/u/36747857?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Rasmus Moorats/>
@@ -832,8 +848,6 @@ make build
             <sub style="font-size:14px"><b>rcursaru</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/renovate-bot>
             <img src=https://avatars.githubusercontent.com/u/25180681?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Mend Renovate/>
@@ -850,9 +864,9 @@ make build
     </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/muzy>
-            <img src=https://avatars.githubusercontent.com/u/321723?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Sebastian Muszytowski/>
+            <img src=https://avatars.githubusercontent.com/u/321723?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Sebastian/>
             <br />
-            <sub style="font-size:14px"><b>Sebastian Muszytowski</b></sub>
+            <sub style="font-size:14px"><b>Sebastian</b></sub>
         </a>
     </td>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
@@ -862,6 +876,8 @@ make build
             <sub style="font-size:14px"><b>Shaanan Cohney</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/6ixfalls>
             <img src=https://avatars.githubusercontent.com/u/23470032?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Six/>
@@ -876,8 +892,6 @@ make build
             <sub style="font-size:14px"><b>Stefan VanBuren</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/sophware>
             <img src=https://avatars.githubusercontent.com/u/41669?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=sophware/>
@@ -906,6 +920,8 @@ make build
             <sub style="font-size:14px"><b>The Gitter Badger</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/tianon>
             <img src=https://avatars.githubusercontent.com/u/161631?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Tianon Gravi/>
@@ -920,8 +936,6 @@ make build
             <sub style="font-size:14px"><b>Till Hoffmann</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/woudsma>
             <img src=https://avatars.githubusercontent.com/u/6162978?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Tjerk Woudsma/>
@@ -950,6 +964,8 @@ make build
             <sub style="font-size:14px"><b>Zachary Newell</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/zekker6>
             <img src=https://avatars.githubusercontent.com/u/1367798?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Zakhar Bessarab/>
@@ -964,8 +980,6 @@ make build
             <sub style="font-size:14px"><b>Zhiyuan Zheng</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/Bpazy>
             <img src=https://avatars.githubusercontent.com/u/9838749?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Ziyuan Han/>
@@ -994,6 +1008,8 @@ make build
             <sub style="font-size:14px"><b>dnaq</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/nning>
             <img src=https://avatars.githubusercontent.com/u/557430?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=henning mueller/>
@@ -1008,8 +1024,6 @@ make build
             <sub style="font-size:14px"><b>ignoramous</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/jimyag>
             <img src=https://avatars.githubusercontent.com/u/69233189?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=jimyag/>
@@ -1038,6 +1052,8 @@ make build
             <sub style="font-size:14px"><b>ma6174</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/manju-rn>
             <img src=https://avatars.githubusercontent.com/u/26291847?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=manju-rn/>
@@ -1052,8 +1068,6 @@ make build
             <sub style="font-size:14px"><b>nicholas-yap</b></sub>
         </a>
     </td>
-</tr>
-<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/pernila>
             <img src=https://avatars.githubusercontent.com/u/12460060?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Tommi Pernila/>
@@ -1082,6 +1096,8 @@ make build
             <sub style="font-size:14px"><b>zy</b></sub>
         </a>
     </td>
+</tr>
+<tr>
     <td align="center" style="word-wrap: break-word; width: 150.0; height: 150.0">
         <a href=https://github.com/atorregrosa-smd>
             <img src=https://avatars.githubusercontent.com/u/78434679?v=4 width="100;"  style="border-radius:50%;align-items:center;justify-content:center;overflow:hidden;padding-top:10px" alt=Àlex Torregrosa/>

cmd/build-docker-img/main.go

@@ -1,47 +0,0 @@
-package main
-
-import (
-	"log"
-
-	"github.com/juanfont/headscale/integration"
-	"github.com/juanfont/headscale/integration/tsic"
-	"github.com/ory/dockertest/v3"
-)
-
-func main() {
-	log.Printf("creating docker pool")
-	pool, err := dockertest.NewPool("")
-	if err != nil {
-		log.Fatalf("could not connect to docker: %s", err)
-	}
-
-	log.Printf("creating docker network")
-	network, err := pool.CreateNetwork("docker-integration-net")
-	if err != nil {
-		log.Fatalf("failed to create or get network: %s", err)
-	}
-
-	for _, version := range integration.AllVersions {
-		log.Printf("creating container image for Tailscale (%s)", version)
-
-		tsClient, err := tsic.New(
-			pool,
-			version,
-			network,
-		)
-		if err != nil {
-			log.Fatalf("failed to create tailscale node: %s", err)
-		}
-
-		err = tsClient.Shutdown()
-		if err != nil {
-			log.Fatalf("failed to shut down container: %s", err)
-		}
-	}
-
-	network.Close()
-	err = pool.RemoveNetwork(network)
-	if err != nil {
-		log.Fatalf("failed to remove network: %s", err)
-	}
-}

cmd/gh-action-integration-generator/main.go

@@ -56,8 +56,11 @@ jobs:
             config-example.yaml
 
       - name: Run {{.Name}}
+        uses: Wandalen/wretry.action@master
         if: steps.changed-files.outputs.any_changed == 'true'
-        run: |
+        with:
+          attempt_limit: 5
+          command: |
             nix develop --command -- docker run \
               --tty --rm \
               --volume ~/.cache/hs-integration-go:/go \
@@ -67,7 +70,6 @@ jobs:
               --volume $PWD/control_logs:/tmp/control \
               golang:1 \
                 go run gotest.tools/gotestsum@latest -- ./... \
-                  -tags ts2019 \
                   -failfast \
                   -timeout 120m \
                   -parallel 1 \

cmd/headscale/cli/api_key.go

@@ -67,7 +67,7 @@ var listAPIKeys = &cobra.Command{
 		}
 
 		if output != "" {
-			SuccessOutput(response.ApiKeys, "", output)
+			SuccessOutput(response.GetApiKeys(), "", output)
 
 			return
 		}
@@ -75,11 +75,11 @@ var listAPIKeys = &cobra.Command{
 		tableData := pterm.TableData{
 			{"ID", "Prefix", "Expiration", "Created"},
 		}
-		for _, key := range response.ApiKeys {
+		for _, key := range response.GetApiKeys() {
 			expiration := "-"
 
 			if key.GetExpiration() != nil {
-				expiration = ColourTime(key.Expiration.AsTime())
+				expiration = ColourTime(key.GetExpiration().AsTime())
 			}
 
 			tableData = append(tableData, []string{
@@ -155,7 +155,7 @@ If you loose a key, create a new one and revoke (expire) the old one.`,
 			return
 		}
 
-		SuccessOutput(response.ApiKey, response.ApiKey, output)
+		SuccessOutput(response.GetApiKey(), response.GetApiKey(), output)
 	},
 }
 

cmd/headscale/cli/debug.go

@@ -4,10 +4,10 @@ import (
 	"fmt"
 
 	v1 "github.com/juanfont/headscale/gen/go/headscale/v1"
-	"github.com/juanfont/headscale/hscontrol/util"
 	"github.com/rs/zerolog/log"
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc/status"
+	"tailscale.com/types/key"
 )
 
 const (
@@ -93,11 +93,13 @@ var createNodeCmd = &cobra.Command{
 
 			return
 		}
-		if !util.NodePublicKeyRegex.Match([]byte(machineKey)) {
+
-			err = errPreAuthKeyMalformed
+		var mkey key.MachinePublic
+		err = mkey.UnmarshalText([]byte(machineKey))
+		if err != nil {
 			ErrorOutput(
 				err,
-				fmt.Sprintf("Error: %s", err),
+				fmt.Sprintf("Failed to parse machine key from flag: %s", err),
 				output,
 			)
 
@@ -133,6 +135,6 @@ var createNodeCmd = &cobra.Command{
 			return
 		}
 
-		SuccessOutput(response.Node, "Node created", output)
+		SuccessOutput(response.GetNode(), "Node created", output)
 	},
 }

cmd/headscale/cli/nodes.go

@@ -152,8 +152,8 @@ var registerNodeCmd = &cobra.Command{
 		}
 
 		SuccessOutput(
-			response.Node,
+			response.GetNode(),
-			fmt.Sprintf("Node %s registered", response.Node.GivenName), output)
+			fmt.Sprintf("Node %s registered", response.GetNode().GetGivenName()), output)
 	},
 }
 
@@ -196,12 +196,12 @@ var listNodesCmd = &cobra.Command{
 		}
 
 		if output != "" {
-			SuccessOutput(response.Nodes, "", output)
+			SuccessOutput(response.GetNodes(), "", output)
 
 			return
 		}
 
-		tableData, err := nodesToPtables(user, showTags, response.Nodes)
+		tableData, err := nodesToPtables(user, showTags, response.GetNodes())
 		if err != nil {
 			ErrorOutput(err, fmt.Sprintf("Error converting to table: %s", err), output)
 
@@ -262,7 +262,7 @@ var expireNodeCmd = &cobra.Command{
 			return
 		}
 
-		SuccessOutput(response.Node, "Node expired", output)
+		SuccessOutput(response.GetNode(), "Node expired", output)
 	},
 }
 
@@ -310,7 +310,7 @@ var renameNodeCmd = &cobra.Command{
 			return
 		}
 
-		SuccessOutput(response.Node, "Node renamed", output)
+		SuccessOutput(response.GetNode(), "Node renamed", output)
 	},
 }
 
@@ -364,7 +364,7 @@ var deleteNodeCmd = &cobra.Command{
 			prompt := &survey.Confirm{
 				Message: fmt.Sprintf(
 					"Do you want to remove the node %s?",
-					getResponse.GetNode().Name,
+					getResponse.GetNode().GetName(),
 				),
 			}
 			err = survey.AskOne(prompt, &confirm)
@@ -473,7 +473,7 @@ var moveNodeCmd = &cobra.Command{
 			return
 		}
 
-		SuccessOutput(moveResponse.Node, "Node moved to another user", output)
+		SuccessOutput(moveResponse.GetNode(), "Node moved to another user", output)
 	},
 }
 
@@ -493,7 +493,7 @@ func nodesToPtables(
 		"Ephemeral",
 		"Last seen",
 		"Expiration",
-		"Online",
+		"Connected",
 		"Expired",
 	}
 	if showTags {
@@ -507,21 +507,21 @@ func nodesToPtables(
 
 	for _, node := range nodes {
 		var ephemeral bool
-		if node.PreAuthKey != nil && node.PreAuthKey.Ephemeral {
+		if node.GetPreAuthKey() != nil && node.GetPreAuthKey().GetEphemeral() {
 			ephemeral = true
 		}
 
 		var lastSeen time.Time
 		var lastSeenTime string
-		if node.LastSeen != nil {
+		if node.GetLastSeen() != nil {
-			lastSeen = node.LastSeen.AsTime()
+			lastSeen = node.GetLastSeen().AsTime()
 			lastSeenTime = lastSeen.Format("2006-01-02 15:04:05")
 		}
 
 		var expiry time.Time
 		var expiryTime string
-		if node.Expiry != nil {
+		if node.GetExpiry() != nil {
-			expiry = node.Expiry.AsTime()
+			expiry = node.GetExpiry().AsTime()
 			expiryTime = expiry.Format("2006-01-02 15:04:05")
 		} else {
 			expiryTime = "N/A"
@@ -529,7 +529,7 @@ func nodesToPtables(
 
 		var machineKey key.MachinePublic
 		err := machineKey.UnmarshalText(
-			[]byte(util.MachinePublicKeyEnsurePrefix(node.MachineKey)),
+			[]byte(node.GetMachineKey()),
 		)
 		if err != nil {
 			machineKey = key.MachinePublic{}
@@ -537,14 +537,14 @@ func nodesToPtables(
 
 		var nodeKey key.NodePublic
 		err = nodeKey.UnmarshalText(
-			[]byte(util.NodePublicKeyEnsurePrefix(node.NodeKey)),
+			[]byte(node.GetNodeKey()),
 		)
 		if err != nil {
 			return nil, err
 		}
 
 		var online string
-		if node.Online {
+		if node.GetOnline() {
 			online = pterm.LightGreen("online")
 		} else {
 			online = pterm.LightRed("offline")
@@ -558,36 +558,36 @@ func nodesToPtables(
 		}
 
 		var forcedTags string
-		for _, tag := range node.ForcedTags {
+		for _, tag := range node.GetForcedTags() {
 			forcedTags += "," + tag
 		}
 		forcedTags = strings.TrimLeft(forcedTags, ",")
 		var invalidTags string
-		for _, tag := range node.InvalidTags {
+		for _, tag := range node.GetInvalidTags() {
-			if !contains(node.ForcedTags, tag) {
+			if !contains(node.GetForcedTags(), tag) {
 				invalidTags += "," + pterm.LightRed(tag)
 			}
 		}
 		invalidTags = strings.TrimLeft(invalidTags, ",")
 		var validTags string
-		for _, tag := range node.ValidTags {
+		for _, tag := range node.GetValidTags() {
-			if !contains(node.ForcedTags, tag) {
+			if !contains(node.GetForcedTags(), tag) {
 				validTags += "," + pterm.LightGreen(tag)
 			}
 		}
 		validTags = strings.TrimLeft(validTags, ",")
 
 		var user string
-		if currentUser == "" || (currentUser == node.User.Name) {
+		if currentUser == "" || (currentUser == node.GetUser().GetName()) {
-			user = pterm.LightMagenta(node.User.Name)
+			user = pterm.LightMagenta(node.GetUser().GetName())
 		} else {
 			// Shared into this user
-			user = pterm.LightYellow(node.User.Name)
+			user = pterm.LightYellow(node.GetUser().GetName())
 		}
 
 		var IPV4Address string
 		var IPV6Address string
-		for _, addr := range node.IpAddresses {
+		for _, addr := range node.GetIpAddresses() {
 			if netip.MustParseAddr(addr).Is4() {
 				IPV4Address = addr
 			} else {
@@ -596,8 +596,8 @@ func nodesToPtables(
 		}
 
 		nodeData := []string{
-			strconv.FormatUint(node.Id, util.Base10),
+			strconv.FormatUint(node.GetId(), util.Base10),
-			node.Name,
+			node.GetName(),
 			node.GetGivenName(),
 			machineKey.ShortString(),
 			nodeKey.ShortString(),

cmd/headscale/cli/preauthkeys.go

@@ -84,7 +84,7 @@ var listPreAuthKeys = &cobra.Command{
 		}
 
 		if output != "" {
-			SuccessOutput(response.PreAuthKeys, "", output)
+			SuccessOutput(response.GetPreAuthKeys(), "", output)
 
 			return
 		}
@@ -101,10 +101,10 @@ var listPreAuthKeys = &cobra.Command{
 				"Tags",
 			},
 		}
-		for _, key := range response.PreAuthKeys {
+		for _, key := range response.GetPreAuthKeys() {
 			expiration := "-"
 			if key.GetExpiration() != nil {
-				expiration = ColourTime(key.Expiration.AsTime())
+				expiration = ColourTime(key.GetExpiration().AsTime())
 			}
 
 			var reusable string
@@ -116,7 +116,7 @@ var listPreAuthKeys = &cobra.Command{
 
 			aclTags := ""
 
-			for _, tag := range key.AclTags {
+			for _, tag := range key.GetAclTags() {
 				aclTags += "," + tag
 			}
 
@@ -214,7 +214,7 @@ var createPreAuthKeyCmd = &cobra.Command{
 			return
 		}
 
-		SuccessOutput(response.PreAuthKey, response.PreAuthKey.Key, output)
+		SuccessOutput(response.GetPreAuthKey(), response.GetPreAuthKey().GetKey(), output)
 	},
 }
 

cmd/headscale/cli/routes.go

@@ -87,12 +87,12 @@ var listRoutesCmd = &cobra.Command{
 			}
 
 			if output != "" {
-				SuccessOutput(response.Routes, "", output)
+				SuccessOutput(response.GetRoutes(), "", output)
 
 				return
 			}
 
-			routes = response.Routes
+			routes = response.GetRoutes()
 		} else {
 			response, err := client.GetNodeRoutes(ctx, &v1.GetNodeRoutesRequest{
 				NodeId: machineID,
@@ -108,12 +108,12 @@ var listRoutesCmd = &cobra.Command{
 			}
 
 			if output != "" {
-				SuccessOutput(response.Routes, "", output)
+				SuccessOutput(response.GetRoutes(), "", output)
 
 				return
 			}
 
-			routes = response.Routes
+			routes = response.GetRoutes()
 		}
 
 		tableData := routesToPtables(routes)
@@ -271,25 +271,25 @@ func routesToPtables(routes []*v1.Route) pterm.TableData {
 
 	for _, route := range routes {
 		var isPrimaryStr string
-		prefix, err := netip.ParsePrefix(route.Prefix)
+		prefix, err := netip.ParsePrefix(route.GetPrefix())
 		if err != nil {
-			log.Printf("Error parsing prefix %s: %s", route.Prefix, err)
+			log.Printf("Error parsing prefix %s: %s", route.GetPrefix(), err)
 
 			continue
 		}
 		if prefix == types.ExitRouteV4 || prefix == types.ExitRouteV6 {
 			isPrimaryStr = "-"
 		} else {
-			isPrimaryStr = strconv.FormatBool(route.IsPrimary)
+			isPrimaryStr = strconv.FormatBool(route.GetIsPrimary())
 		}
 
 		tableData = append(tableData,
 			[]string{
-				strconv.FormatUint(route.Id, Base10),
+				strconv.FormatUint(route.GetId(), Base10),
-				route.Node.GivenName,
+				route.GetNode().GetGivenName(),
-				route.Prefix,
+				route.GetPrefix(),
-				strconv.FormatBool(route.Advertised),
+				strconv.FormatBool(route.GetAdvertised()),
-				strconv.FormatBool(route.Enabled),
+				strconv.FormatBool(route.GetEnabled()),
 				isPrimaryStr,
 			})
 	}

cmd/headscale/cli/users.go

@@ -67,7 +67,7 @@ var createUserCmd = &cobra.Command{
 			return
 		}
 
-		SuccessOutput(response.User, "User created", output)
+		SuccessOutput(response.GetUser(), "User created", output)
 	},
 }
 
@@ -169,7 +169,7 @@ var listUsersCmd = &cobra.Command{
 		}
 
 		if output != "" {
-			SuccessOutput(response.Users, "", output)
+			SuccessOutput(response.GetUsers(), "", output)
 
 			return
 		}
@@ -236,6 +236,6 @@ var renameUserCmd = &cobra.Command{
 			return
 		}
 
-		SuccessOutput(response.User, "User renamed", output)
+		SuccessOutput(response.GetUser(), "User renamed", output)
 	},
 }

config-example.yaml

@@ -40,19 +40,12 @@ grpc_listen_addr: 127.0.0.1:50443
 # are doing.
 grpc_allow_insecure: false
 
-# Private key used to encrypt the traffic between headscale
-# and Tailscale clients.
-# The private key file will be autogenerated if it's missing.
-#
-private_key_path: /var/lib/headscale/private.key
-
 # The Noise section includes specific configuration for the
 # TS2021 Noise protocol
 noise:
   # The Noise private key is used to encrypt the
   # traffic between headscale and Tailscale clients when
-  # using the new Noise-based protocol. It must be different
+  # using the new Noise-based protocol.
-  # from the legacy private key.
   private_key_path: /var/lib/headscale/noise_private.key
 
 # List of IP prefixes to allocate tailaddresses from.
@@ -95,6 +88,12 @@ derp:
     # For more details on how this works, check this great article: https://tailscale.com/blog/how-tailscale-works/
     stun_listen_addr: "0.0.0.0:3478"
 
+    # Private key used to encrypt the traffic between headscale DERP
+    # and Tailscale clients.
+    # The private key file will be autogenerated if it's missing.
+    #
+    private_key_path: /var/lib/headscale/derp_server_private.key
+
   # List of externally available DERP maps encoded in JSON
   urls:
     - https://controlplane.tailscale.com/derpmap/default

docs/requirements.txt

@@ -0,0 +1,5 @@
+cairosvg~=2.7.1
+mkdocs-material~=9.4.14
+mkdocs-minify-plugin~=0.7.1
+pillow~=10.1.0
+

docs/running-headscale-container.md

@@ -28,7 +28,7 @@ cd ./headscale
 touch ./config/db.sqlite
 ```
 
-3. **(Strongly Recommended)** Download a copy of the [example configuration][config-example.yaml](https://github.com/juanfont/headscale/blob/main/config-example.yaml) from the headscale repository.
+3. **(Strongly Recommended)** Download a copy of the [example configuration](https://github.com/juanfont/headscale/blob/main/config-example.yaml) from the headscale repository.
 
 Using wget:
 

flake.lock

@@ -5,11 +5,11 @@
         "systems": "systems"
       },
       "locked": {
-        "lastModified": 1694529238,
+        "lastModified": 1701680307,
-        "narHash": "sha256-zsNZZGTGnMOf9YpHKJqMSsa0dXbfmxeoJ7xHlrt+xmY=",
+        "narHash": "sha256-kAuep2h5ajznlPMD9rnQyffWG8EM/C73lejGofXvdM8=",
         "owner": "numtide",
         "repo": "flake-utils",
-        "rev": "ff7b65b44d01cf9ba6a71320833626af21126384",
+        "rev": "4022d587cbbfd70fe950c1e2083a02621806a725",
         "type": "github"
       },
       "original": {
@@ -20,11 +20,11 @@
     },
     "nixpkgs": {
       "locked": {
-        "lastModified": 1699186365,
+        "lastModified": 1701998057,
-        "narHash": "sha256-Pxrw5U8mBsL3NlrJ6q1KK1crzvSUcdfwb9083sKDrcU=",
+        "narHash": "sha256-gAJGhcTO9cso7XDfAScXUlPcva427AUT2q02qrmXPdo=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "a0b3b06b7a82c965ae0bb1d59f6e386fe755001d",
+        "rev": "09dc04054ba2ff1f861357d0e7e76d021b273cd7",
         "type": "github"
       },
       "original": {

flake.nix

@@ -26,14 +26,12 @@
           version = headscaleVersion;
           src = pkgs.lib.cleanSource self;
 
-          tags = ["ts2019"];
-
           # Only run unit tests when testing a build
           checkFlags = ["-short"];
 
           # When updating go.mod or go.sum, a new sha will need to be calculated,
           # update this if you have a mismatch after doing a change to thos files.
-          vendorSha256 = "sha256-Q6eySc8lXYhkWka7Y+qOM6viv7QhdjFZDX8PttaLfr4=";
+          vendorHash = "sha256-8x4RKaS8vnBYTPlvQTkDKWIAJOgPF99hvPiuRyTMrA8=";
 
           ldflags = ["-s" "-w" "-X github.com/juanfont/headscale/cmd/headscale/cli.Version=v${version}"];
         };
@@ -49,7 +47,7 @@
             sha256 = "sha256-2K9KAg8iSubiTbujyFGN3yggrL+EDyeUCs9OOta/19A=";
           };
 
-          vendorSha256 = "sha256-rxYuzn4ezAxaeDhxd8qdOzt+CKYIh03A9zKNdzILq18=";
+          vendorHash = "sha256-rxYuzn4ezAxaeDhxd8qdOzt+CKYIh03A9zKNdzILq18=";
 
           nativeBuildInputs = [pkgs.installShellFiles];
         };
@@ -71,7 +69,7 @@
             sha256 = "sha256-lnNdsDCpeSHtl2lC1IhUw11t3cnGF+37qSM7HDvKLls=";
           };
 
-          vendorSha256 = "sha256-dGdnDuRbwg8fU7uB5GaHEWa/zI3w06onqjturvooJQA=";
+          vendorHash = "sha256-dGdnDuRbwg8fU7uB5GaHEWa/zI3w06onqjturvooJQA=";
 
           nativeBuildInputs = [pkgs.installShellFiles];
 
@@ -129,15 +127,7 @@
         buildInputs = devDeps;
 
         shellHook = ''
-          export GOFLAGS=-tags="ts2019"
           export PATH="$PWD/result/bin:$PATH"
-
-          mkdir -p ./ignored
-          export HEADSCALE_PRIVATE_KEY_PATH="./ignored/private.key"
-          export HEADSCALE_NOISE_PRIVATE_KEY_PATH="./ignored/noise_private.key"
-          export HEADSCALE_DB_PATH="./ignored/db.sqlite"
-          export HEADSCALE_TLS_LETSENCRYPT_CACHE_DIR="./ignored/cache"
-          export HEADSCALE_UNIX_SOCKET="./ignored/headscale.sock"
         '';
       };
 

go.mod

@@ -1,137 +1,181 @@
 module github.com/juanfont/headscale
 
-go 1.21
+go 1.21.0
 
-toolchain go1.21.1
+toolchain go1.21.4
 
 require (
-	github.com/AlecAivazis/survey/v2 v2.3.6
+	github.com/AlecAivazis/survey/v2 v2.3.7
-	github.com/coreos/go-oidc/v3 v3.5.0
+	github.com/coreos/go-oidc/v3 v3.8.0
-	github.com/davecgh/go-spew v1.1.1
+	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc
-	github.com/deckarep/golang-set/v2 v2.3.0
+	github.com/deckarep/golang-set/v2 v2.4.0
 	github.com/efekarakus/termcolor v1.0.1
-	github.com/glebarez/sqlite v1.7.0
+	github.com/glebarez/sqlite v1.10.0
+	github.com/go-gormigrate/gormigrate/v2 v2.1.1
 	github.com/gofrs/uuid/v5 v5.0.0
-	github.com/google/go-cmp v0.5.9
+	github.com/google/go-cmp v0.6.0
-	github.com/gorilla/mux v1.8.0
+	github.com/gorilla/mux v1.8.1
 	github.com/grpc-ecosystem/go-grpc-middleware v1.4.0
-	github.com/grpc-ecosystem/grpc-gateway/v2 v2.15.2
+	github.com/grpc-ecosystem/grpc-gateway/v2 v2.18.1
-	github.com/klauspost/compress v1.16.7
+	github.com/klauspost/compress v1.17.3
 	github.com/oauth2-proxy/mockoidc v0.0.0-20220308204021-b9169deeb282
-	github.com/ory/dockertest/v3 v3.9.1
+	github.com/ory/dockertest/v3 v3.10.0
 	github.com/patrickmn/go-cache v2.1.0+incompatible
 	github.com/philip-bui/grpc-zerolog v1.0.1
 	github.com/pkg/profile v1.7.0
-	github.com/prometheus/client_golang v1.15.1
+	github.com/prometheus/client_golang v1.17.0
-	github.com/prometheus/common v0.42.0
+	github.com/prometheus/common v0.45.0
-	github.com/pterm/pterm v0.12.58
+	github.com/pterm/pterm v0.12.71
-	github.com/puzpuzpuz/xsync/v2 v2.4.0
+	github.com/puzpuzpuz/xsync/v3 v3.0.2
-	github.com/rs/zerolog v1.29.0
+	github.com/rs/zerolog v1.31.0
 	github.com/samber/lo v1.38.1
-	github.com/spf13/cobra v1.7.0
+	github.com/spf13/cobra v1.8.0
-	github.com/spf13/viper v1.16.0
+	github.com/spf13/viper v1.17.0
 	github.com/stretchr/testify v1.8.4
 	github.com/tailscale/hujson v0.0.0-20221223112325-20486734a56a
+	github.com/tailscale/tailsql v0.0.0-20231216172832-51483e0c711b
 	github.com/tcnksm/go-latest v0.0.0-20170313132115-e3007ae9052e
-	go4.org/netipx v0.0.0-20230728180743-ad4cb58a6516
+	go4.org/netipx v0.0.0-20230824141953-6213f710f925
-	golang.org/x/crypto v0.12.0
+	golang.org/x/crypto v0.16.0
-	golang.org/x/net v0.14.0
+	golang.org/x/exp v0.0.0-20231127185646-65229373498e
-	golang.org/x/oauth2 v0.7.0
+	golang.org/x/net v0.19.0
-	golang.org/x/sync v0.2.0
+	golang.org/x/oauth2 v0.15.0
-	google.golang.org/genproto v0.0.0-20230410155749-daa745c078e1
+	golang.org/x/sync v0.5.0
-	google.golang.org/grpc v1.55.0
+	google.golang.org/genproto/googleapis/api v0.0.0-20231127180814-3a041ad873d4
-	google.golang.org/protobuf v1.30.0
+	google.golang.org/grpc v1.59.0
+	google.golang.org/protobuf v1.31.0
 	gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c
 	gopkg.in/yaml.v3 v3.0.1
-	gorm.io/driver/postgres v1.4.8
+	gorm.io/driver/postgres v1.5.4
-	gorm.io/gorm v1.24.6
+	gorm.io/gorm v1.25.5
-	tailscale.com v1.50.0
+	tailscale.com v1.56.1
 )
 
 require (
-	atomicgo.dev/cursor v0.1.1 // indirect
+	atomicgo.dev/cursor v0.2.0 // indirect
 	atomicgo.dev/keyboard v0.2.9 // indirect
+	atomicgo.dev/schedule v0.1.0 // indirect
 	filippo.io/edwards25519 v1.0.0 // indirect
 	github.com/Azure/go-ansiterm v0.0.0-20230124172434-306776ec8161 // indirect
 	github.com/Microsoft/go-winio v0.6.1 // indirect
 	github.com/Nvveen/Gotty v0.0.0-20120604004816-cd527374f1e5 // indirect
 	github.com/akutz/memconn v0.1.0 // indirect
-	github.com/alexbrainman/sspi v0.0.0-20210105120005-909beea2cc74 // indirect
+	github.com/alexbrainman/sspi v0.0.0-20231016080023-1a75b4708caa // indirect
+	github.com/aws/aws-sdk-go-v2 v1.21.0 // indirect
+	github.com/aws/aws-sdk-go-v2/config v1.18.42 // indirect
+	github.com/aws/aws-sdk-go-v2/credentials v1.13.40 // indirect
+	github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.13.11 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/configsources v1.1.41 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.4.35 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/ini v1.3.43 // indirect
+	github.com/aws/aws-sdk-go-v2/service/internal/presigned-url v1.9.35 // indirect
+	github.com/aws/aws-sdk-go-v2/service/ssm v1.38.0 // indirect
+	github.com/aws/aws-sdk-go-v2/service/sso v1.14.1 // indirect
+	github.com/aws/aws-sdk-go-v2/service/ssooidc v1.17.1 // indirect
+	github.com/aws/aws-sdk-go-v2/service/sts v1.22.0 // indirect
+	github.com/aws/smithy-go v1.14.2 // indirect
 	github.com/beorn7/perks v1.0.1 // indirect
-	github.com/cenkalti/backoff/v4 v4.2.0 // indirect
+	github.com/cenkalti/backoff/v4 v4.2.1 // indirect
 	github.com/cespare/xxhash/v2 v2.2.0 // indirect
 	github.com/containerd/console v1.0.3 // indirect
-	github.com/containerd/continuity v0.3.0 // indirect
+	github.com/containerd/continuity v0.4.3 // indirect
-	github.com/coreos/go-iptables v0.6.0 // indirect
+	github.com/coreos/go-iptables v0.7.0 // indirect
-	github.com/dblohm7/wingoes v0.0.0-20230821191801-fc76608aecf0 // indirect
+	github.com/coreos/go-systemd/v22 v22.5.0 // indirect
-	github.com/docker/cli v23.0.5+incompatible // indirect
+	github.com/dblohm7/wingoes v0.0.0-20231025182615-65d8b4b5428f // indirect
-	github.com/docker/docker v24.0.4+incompatible // indirect
+	github.com/digitalocean/go-smbios v0.0.0-20180907143718-390a4f403a8e // indirect
+	github.com/docker/cli v24.0.7+incompatible // indirect
+	github.com/docker/docker v24.0.7+incompatible // indirect
 	github.com/docker/go-connections v0.4.0 // indirect
 	github.com/docker/go-units v0.5.0 // indirect
 	github.com/dustin/go-humanize v1.0.1 // indirect
 	github.com/felixge/fgprof v0.9.3 // indirect
-	github.com/fsnotify/fsnotify v1.6.0 // indirect
+	github.com/fsnotify/fsnotify v1.7.0 // indirect
-	github.com/fxamacker/cbor/v2 v2.4.0 // indirect
+	github.com/fxamacker/cbor/v2 v2.5.0 // indirect
-	github.com/glebarez/go-sqlite v1.20.3 // indirect
+	github.com/glebarez/go-sqlite v1.21.2 // indirect
-	github.com/go-jose/go-jose/v3 v3.0.0 // indirect
+	github.com/go-jose/go-jose/v3 v3.0.1 // indirect
+	github.com/go-ole/go-ole v1.3.0 // indirect
+	github.com/godbus/dbus/v5 v5.1.1-0.20230522191255-76236955d466 // indirect
 	github.com/gogo/protobuf v1.3.2 // indirect
 	github.com/golang-jwt/jwt v3.2.2+incompatible // indirect
 	github.com/golang/groupcache v0.0.0-20210331224755-41bb18bfe9da // indirect
 	github.com/golang/protobuf v1.5.3 // indirect
+	github.com/google/btree v1.1.2 // indirect
 	github.com/google/go-github v17.0.0+incompatible // indirect
 	github.com/google/go-querystring v1.1.0 // indirect
 	github.com/google/nftables v0.1.1-0.20230115205135-9aa6fdf5a28c // indirect
-	github.com/google/pprof v0.0.0-20221118152302-e6195bd50e26 // indirect
+	github.com/google/pprof v0.0.0-20231127191134-f3a68a39ae15 // indirect
 	github.com/google/shlex v0.0.0-20191202100458-e7afc7fbc510 // indirect
-	github.com/google/uuid v1.3.0 // indirect
+	github.com/google/uuid v1.4.0 // indirect
-	github.com/gookit/color v1.5.3 // indirect
+	github.com/gookit/color v1.5.4 // indirect
+	github.com/gorilla/csrf v1.7.1 // indirect
+	github.com/gorilla/securecookie v1.1.1 // indirect
 	github.com/hashicorp/go-version v1.6.0 // indirect
 	github.com/hashicorp/hcl v1.0.0 // indirect
 	github.com/hdevalence/ed25519consensus v0.1.0 // indirect
+	github.com/illarion/gonotify v1.0.1 // indirect
 	github.com/imdario/mergo v0.3.16 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
+	github.com/insomniacslk/dhcp v0.0.0-20230908212754-65c27093e38a // indirect
 	github.com/jackc/pgpassfile v1.0.0 // indirect
 	github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a // indirect
-	github.com/jackc/pgx/v5 v5.3.0 // indirect
+	github.com/jackc/pgx/v5 v5.5.0 // indirect
+	github.com/jackc/puddle/v2 v2.2.1 // indirect
 	github.com/jinzhu/inflection v1.0.0 // indirect
 	github.com/jinzhu/now v1.1.5 // indirect
+	github.com/jmespath/go-jmespath v0.4.0 // indirect
 	github.com/josharian/native v1.1.1-0.20230202152459-5c7d0dd6ab86 // indirect
-	github.com/jsimonetti/rtnetlink v1.3.2 // indirect
+	github.com/jsimonetti/rtnetlink v1.4.0 // indirect
 	github.com/kballard/go-shellquote v0.0.0-20180428030007-95032a82bc51 // indirect
+	github.com/kortschak/wol v0.0.0-20200729010619-da482cc4850a // indirect
 	github.com/kr/pretty v0.3.1 // indirect
 	github.com/kr/text v0.2.0 // indirect
 	github.com/lib/pq v1.10.7 // indirect
-	github.com/lithammer/fuzzysearch v1.1.5 // indirect
+	github.com/lithammer/fuzzysearch v1.1.8 // indirect
 	github.com/magiconair/properties v1.8.7 // indirect
 	github.com/mattn/go-colorable v0.1.13 // indirect
-	github.com/mattn/go-isatty v0.0.18 // indirect
+	github.com/mattn/go-isatty v0.0.20 // indirect
-	github.com/mattn/go-runewidth v0.0.14 // indirect
+	github.com/mattn/go-runewidth v0.0.15 // indirect
-	github.com/matttproud/golang_protobuf_extensions v1.0.4 // indirect
+	github.com/matttproud/golang_protobuf_extensions/v2 v2.0.0 // indirect
+	github.com/mdlayher/genetlink v1.3.2 // indirect
 	github.com/mdlayher/netlink v1.7.2 // indirect
-	github.com/mdlayher/socket v0.4.1 // indirect
+	github.com/mdlayher/sdnotify v1.0.0 // indirect
+	github.com/mdlayher/socket v0.5.0 // indirect
 	github.com/mgutz/ansi v0.0.0-20200706080929-d51e80ef957d // indirect
-	github.com/miekg/dns v1.1.55 // indirect
+	github.com/miekg/dns v1.1.57 // indirect
 	github.com/mitchellh/go-ps v1.0.0 // indirect
 	github.com/mitchellh/mapstructure v1.5.0 // indirect
-	github.com/moby/term v0.0.0-20221205130635-1aeaba878587 // indirect
+	github.com/moby/term v0.5.0 // indirect
 	github.com/opencontainers/go-digest v1.0.0 // indirect
-	github.com/opencontainers/image-spec v1.1.0-rc3 // indirect
+	github.com/opencontainers/image-spec v1.1.0-rc5 // indirect
-	github.com/opencontainers/runc v1.1.4 // indirect
+	github.com/opencontainers/runc v1.1.10 // indirect
-	github.com/pelletier/go-toml/v2 v2.0.8 // indirect
+	github.com/pelletier/go-toml/v2 v2.1.0 // indirect
+	github.com/pierrec/lz4/v4 v4.1.18 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
-	github.com/pmezard/go-difflib v1.0.0 // indirect
+	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
-	github.com/prometheus/client_model v0.4.0 // indirect
+	github.com/prometheus/client_model v0.5.0 // indirect
-	github.com/prometheus/procfs v0.9.0 // indirect
+	github.com/prometheus/procfs v0.12.0 // indirect
 	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
 	github.com/rivo/uniseg v0.4.4 // indirect
-	github.com/rogpeppe/go-internal v1.10.0 // indirect
+	github.com/rogpeppe/go-internal v1.11.0 // indirect
-	github.com/sirupsen/logrus v1.9.0 // indirect
+	github.com/safchain/ethtool v0.3.0 // indirect
-	github.com/spf13/afero v1.9.5 // indirect
+	github.com/sagikazarmark/locafero v0.4.0 // indirect
-	github.com/spf13/cast v1.5.1 // indirect
+	github.com/sagikazarmark/slog-shim v0.1.0 // indirect
-	github.com/spf13/jwalterweatherman v1.1.0 // indirect
+	github.com/sirupsen/logrus v1.9.3 // indirect
+	github.com/sourcegraph/conc v0.3.0 // indirect
+	github.com/spf13/afero v1.11.0 // indirect
+	github.com/spf13/cast v1.6.0 // indirect
 	github.com/spf13/pflag v1.0.5 // indirect
-	github.com/subosito/gotenv v1.4.2 // indirect
+	github.com/subosito/gotenv v1.6.0 // indirect
+	github.com/tailscale/certstore v0.1.1-0.20231202035212-d3fa0460f47e // indirect
+	github.com/tailscale/go-winio v0.0.0-20231025203758-c4f33415bf55 // indirect
+	github.com/tailscale/golang-x-crypto v0.0.0-20230713185742-f0b76a10a08e // indirect
+	github.com/tailscale/goupnp v1.0.1-0.20210804011211-c64d0f06ea05 // indirect
 	github.com/tailscale/netlink v1.1.1-0.20211101221916-cabfb018fe85 // indirect
+	github.com/tailscale/setec v0.0.0-20230926024544-07dde05889e7 // indirect
+	github.com/tailscale/web-client-prebuilt v0.0.0-20231213172531-a4fa669015b2 // indirect
+	github.com/tailscale/wireguard-go v0.0.0-20231121184858-cc193a0b3272 // indirect
+	github.com/tcnksm/go-httpstat v0.2.0 // indirect
+	github.com/u-root/uio v0.0.0-20230305220412-3e8cd9d6bf63 // indirect
 	github.com/vishvananda/netlink v1.2.1-beta.2 // indirect
 	github.com/vishvananda/netns v0.0.4 // indirect
 	github.com/x448/float16 v0.8.4 // indirect
@@ -139,23 +183,27 @@ require (
 	github.com/xeipuuv/gojsonreference v0.0.0-20180127040603-bd5ef7bd5415 // indirect
 	github.com/xeipuuv/gojsonschema v1.2.0 // indirect
 	github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e // indirect
+	go.uber.org/multierr v1.11.0 // indirect
 	go4.org/mem v0.0.0-20220726221520-4f986261bf13 // indirect
-	golang.org/x/exp v0.0.0-20230725093048-515e97ebf090 // indirect
+	golang.org/x/mod v0.14.0 // indirect
-	golang.org/x/mod v0.11.0 // indirect
+	golang.org/x/sys v0.15.0 // indirect
-	golang.org/x/sys v0.11.0 // indirect
+	golang.org/x/term v0.15.0 // indirect
-	golang.org/x/term v0.11.0 // indirect
+	golang.org/x/text v0.14.0 // indirect
-	golang.org/x/text v0.12.0 // indirect
+	golang.org/x/time v0.5.0 // indirect
-	golang.org/x/time v0.3.0 // indirect
+	golang.org/x/tools v0.16.0 // indirect
-	golang.org/x/tools v0.9.1 // indirect
+	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect
 	golang.zx2c4.com/wireguard/windows v0.5.3 // indirect
-	google.golang.org/appengine v1.6.7 // indirect
+	google.golang.org/appengine v1.6.8 // indirect
+	google.golang.org/genproto v0.0.0-20231127180814-3a041ad873d4 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20231127180814-3a041ad873d4 // indirect
 	gopkg.in/ini.v1 v1.67.0 // indirect
 	gopkg.in/square/go-jose.v2 v2.6.0 // indirect
 	gopkg.in/yaml.v2 v2.4.0 // indirect
-	gotest.tools/v3 v3.4.0 // indirect
+	gvisor.dev/gvisor v0.0.0-20230928000133-4fe30062272c // indirect
-	modernc.org/libc v1.22.2 // indirect
+	inet.af/peercred v0.0.0-20210906144145-0893ea02156a // indirect
-	modernc.org/mathutil v1.5.0 // indirect
+	modernc.org/libc v1.34.11 // indirect
-	modernc.org/memory v1.5.0 // indirect
+	modernc.org/mathutil v1.6.0 // indirect
-	modernc.org/sqlite v1.20.3 // indirect
+	modernc.org/memory v1.7.2 // indirect
-	nhooyr.io/websocket v1.8.7 // indirect
+	modernc.org/sqlite v1.28.0 // indirect
+	nhooyr.io/websocket v1.8.10 // indirect
 )

hscontrol/app.go

@@ -48,6 +48,7 @@ import (
 	"google.golang.org/grpc/peer"
 	"google.golang.org/grpc/reflection"
 	"google.golang.org/grpc/status"
+	"tailscale.com/envknob"
 	"tailscale.com/tailcfg"
 	"tailscale.com/types/dnstype"
 	"tailscale.com/types/key"
@@ -59,6 +60,9 @@ var (
 	errUnsupportedLetsEncryptChallengeType = errors.New(
 		"unknown value for Lets Encrypt challenge type",
 	)
+	errEmptyInitialDERPMap = errors.New(
+		"initial DERPMap is empty, Headscale requries at least one entry",
+	)
 )
 
 const (
@@ -77,7 +81,6 @@ type Headscale struct {
 	dbString        string
 	dbType          string
 	dbDebug         bool
-	privateKey2019  *key.MachinePrivate
 	noisePrivateKey *key.MachinePrivate
 
 	DERPMap    *tailcfg.DERPMap
@@ -96,26 +99,23 @@ type Headscale struct {
 	pollNetMapStreamWG sync.WaitGroup
 }
 
+var (
+	profilingEnabled = envknob.Bool("HEADSCALE_PROFILING_ENABLED")
+	tailsqlEnabled   = envknob.Bool("HEADSCALE_DEBUG_TAILSQL_ENABLED")
+	tailsqlStateDir  = envknob.String("HEADSCALE_DEBUG_TAILSQL_STATE_DIR")
+	tailsqlTSKey     = envknob.String("TS_AUTHKEY")
+)
+
 func NewHeadscale(cfg *types.Config) (*Headscale, error) {
-	if _, enableProfile := os.LookupEnv("HEADSCALE_PROFILING_ENABLED"); enableProfile {
+	if profilingEnabled {
 		runtime.SetBlockProfileRate(1)
 	}
 
-	privateKey, err := readOrCreatePrivateKey(cfg.PrivateKeyPath)
-	if err != nil {
-		return nil, fmt.Errorf("failed to read or create private key: %w", err)
-	}
-
-	// TS2021 requires to have a different key from the legacy protocol.
 	noisePrivateKey, err := readOrCreatePrivateKey(cfg.NoisePrivateKeyPath)
 	if err != nil {
 		return nil, fmt.Errorf("failed to read or create Noise protocol private key: %w", err)
 	}
 
-	if privateKey.Equal(*noisePrivateKey) {
-		return nil, fmt.Errorf("private key and noise private key are the same: %w", err)
-	}
-
 	var dbString string
 	switch cfg.DBtype {
 	case db.Postgres:
@@ -156,7 +156,6 @@ func NewHeadscale(cfg *types.Config) (*Headscale, error) {
 		cfg:                cfg,
 		dbType:             cfg.DBtype,
 		dbString:           dbString,
-		privateKey2019:     privateKey,
 		noisePrivateKey:    noisePrivateKey,
 		registrationCache:  registrationCache,
 		pollNetMapStreamWG: sync.WaitGroup{},
@@ -199,10 +198,21 @@ func NewHeadscale(cfg *types.Config) (*Headscale, error) {
 	}
 
 	if cfg.DERP.ServerEnabled {
-		// TODO(kradalby): replace this key with a dedicated DERP key.
+		derpServerKey, err := readOrCreatePrivateKey(cfg.DERP.ServerPrivateKeyPath)
+		if err != nil {
+			return nil, fmt.Errorf("failed to read or create DERP server private key: %w", err)
+		}
+
+		if derpServerKey.Equal(*noisePrivateKey) {
+			return nil, fmt.Errorf(
+				"DERP server private key and noise private key are the same: %w",
+				err,
+			)
+		}
+
 		embeddedDERPServer, err := derpServer.NewDERPServer(
 			cfg.ServerURL,
-			key.NodePrivate(*privateKey),
+			key.NodePrivate(*derpServerKey),
 			&cfg.DERP,
 		)
 		if err != nil {
@@ -263,20 +273,13 @@ func (h *Headscale) scheduledDERPMapUpdateWorker(cancelChan <-chan struct{}) {
 				h.DERPMap.Regions[region.RegionID] = &region
 			}
 
-			h.nodeNotifier.NotifyAll(types.StateUpdate{
+			stateUpdate := types.StateUpdate{
 				Type:    types.StateDERPUpdated,
-				DERPMap: *h.DERPMap,
+				DERPMap: h.DERPMap,
-			})
+			}
-		}
+			if stateUpdate.Valid() {
-	}
+				h.nodeNotifier.NotifyAll(stateUpdate)
-}
+			}
-
-func (h *Headscale) failoverSubnetRoutes(milliSeconds int64) {
-	ticker := time.NewTicker(time.Duration(milliSeconds) * time.Millisecond)
-	for range ticker.C {
-		err := h.db.HandlePrimarySubnetFailover()
-		if err != nil {
-			log.Error().Err(err).Msg("failed to handle primary subnet failover")
 		}
 	}
 }
@@ -449,10 +452,9 @@ func (h *Headscale) createRouter(grpcMux *grpcRuntime.ServeMux) *mux.Router {
 
 	router.HandleFunc("/health", h.HealthHandler).Methods(http.MethodGet)
 	router.HandleFunc("/key", h.KeyHandler).Methods(http.MethodGet)
-	router.HandleFunc("/register/{nkey}", h.RegisterWebAPI).Methods(http.MethodGet)
+	router.HandleFunc("/register/{mkey}", h.RegisterWebAPI).Methods(http.MethodGet)
-	h.addLegacyHandlers(router)
 
-	router.HandleFunc("/oidc/register/{nkey}", h.RegisterOIDC).Methods(http.MethodGet)
+	router.HandleFunc("/oidc/register/{mkey}", h.RegisterOIDC).Methods(http.MethodGet)
 	router.HandleFunc("/oidc/callback", h.OIDCCallback).Methods(http.MethodGet)
 	router.HandleFunc("/apple", h.AppleConfigMessage).Methods(http.MethodGet)
 	router.HandleFunc("/apple/{platform}", h.ApplePlatformConfig).
@@ -510,13 +512,15 @@ func (h *Headscale) Serve() error {
 		go h.scheduledDERPMapUpdateWorker(derpMapCancelChannel)
 	}
 
+	if len(h.DERPMap.Regions) == 0 {
+		return errEmptyInitialDERPMap
+	}
+
 	// TODO(kradalby): These should have cancel channels and be cleaned
 	// up on shutdown.
 	go h.expireEphemeralNodes(updateInterval)
 	go h.expireExpiredMachines(updateInterval)
 
-	go h.failoverSubnetRoutes(updateInterval)
-
 	if zl.GlobalLevel() == zl.TraceLevel {
 		zerolog.RespLog = true
 	} else {
@@ -572,7 +576,10 @@ func (h *Headscale) Serve() error {
 	}
 
 	// Start the local gRPC server without TLS and without authentication
-	grpcSocket := grpc.NewServer(zerolog.UnaryInterceptor())
+	grpcSocket := grpc.NewServer(
+	// Uncomment to debug grpc communication.
+	// zerolog.UnaryInterceptor(),
+	)
 
 	v1.RegisterHeadscaleServiceServer(grpcSocket, newHeadscaleV1APIServer(h))
 	reflection.Register(grpcSocket)
@@ -612,7 +619,8 @@ func (h *Headscale) Serve() error {
 			grpc.UnaryInterceptor(
 				grpcMiddleware.ChainUnaryServer(
 					h.grpcAuthenticationInterceptor,
-					zerolog.NewUnaryServerInterceptor(),
+					// Uncomment to debug grpc communication.
+					// zerolog.NewUnaryServerInterceptor(),
 				),
 			),
 		}
@@ -698,6 +706,18 @@ func (h *Headscale) Serve() error {
 	log.Info().
 		Msgf("listening and serving metrics on: %s", h.cfg.MetricsAddr)
 
+	var tailsqlContext context.Context
+	if tailsqlEnabled {
+		if h.cfg.DBtype != db.Sqlite {
+			log.Fatal().Str("type", h.cfg.DBtype).Msgf("tailsql only support %q", db.Sqlite)
+		}
+		if tailsqlTSKey == "" {
+			log.Fatal().Msg("tailsql requires TS_AUTHKEY to be set")
+		}
+		tailsqlContext = context.Background()
+		go runTailSQLService(ctx, util.TSLogfWrapper(), tailsqlStateDir, h.cfg.DBpath)
+	}
+
 	// Handle common process-killing signals so we can gracefully shut down:
 	h.shutdownChan = make(chan struct{})
 	sigc := make(chan os.Signal, 1)
@@ -763,6 +783,10 @@ func (h *Headscale) Serve() error {
 					grpcListener.Close()
 				}
 
+				if tailsqlContext != nil {
+					tailsqlContext.Done()
+				}
+
 				// Close network listeners
 				promHTTPListener.Close()
 				httpListener.Close()
@@ -900,7 +924,8 @@ func readOrCreatePrivateKey(path string) (*key.MachinePrivate, error) {
 		err = os.WriteFile(path, machineKeyStr, privateKeyFileMode)
 		if err != nil {
 			return nil, fmt.Errorf(
-				"failed to save private key to disk: %w",
+				"failed to save private key to disk at path %q: %w",
+				path,
 				err,
 			)
 		}
@@ -911,16 +936,9 @@ func readOrCreatePrivateKey(path string) (*key.MachinePrivate, error) {
 	}
 
 	trimmedPrivateKey := strings.TrimSpace(string(privateKey))
-	privateKeyEnsurePrefix := util.PrivateKeyEnsurePrefix(trimmedPrivateKey)
 
 	var machineKey key.MachinePrivate
-	if err = machineKey.UnmarshalText([]byte(privateKeyEnsurePrefix)); err != nil {
+	if err = machineKey.UnmarshalText([]byte(trimmedPrivateKey)); err != nil {
-		log.Info().
-			Str("path", path).
-			Msg("This might be due to a legacy (headscale pre-0.12) private key. " +
-				"If the key is in WireGuard format, delete the key and restart headscale. " +
-				"A new key will automatically be generated. All Tailscale clients will have to be restarted")
-
 		return nil, fmt.Errorf("failed to parse private key: %w", err)
 	}
 

hscontrol/auth.go

@@ -1,13 +1,13 @@
 package hscontrol
 
 import (
+	"encoding/json"
 	"errors"
 	"fmt"
 	"net/http"
 	"strings"
 	"time"
 
-	"github.com/juanfont/headscale/hscontrol/mapper"
 	"github.com/juanfont/headscale/hscontrol/types"
 	"github.com/juanfont/headscale/hscontrol/util"
 	"github.com/rs/zerolog/log"
@@ -16,22 +16,62 @@ import (
 	"tailscale.com/types/key"
 )
 
-// handleRegister is the common logic for registering a client in the legacy and Noise protocols
+func logAuthFunc(
-//
+	registerRequest tailcfg.RegisterRequest,
-// When using Noise, the machineKey is Zero.
+	machineKey key.MachinePublic,
+) (func(string), func(string), func(error, string)) {
+	return func(msg string) {
+			log.Info().
+				Caller().
+				Str("machine_key", machineKey.ShortString()).
+				Str("node_key", registerRequest.NodeKey.ShortString()).
+				Str("node_key_old", registerRequest.OldNodeKey.ShortString()).
+				Str("node", registerRequest.Hostinfo.Hostname).
+				Str("followup", registerRequest.Followup).
+				Time("expiry", registerRequest.Expiry).
+				Msg(msg)
+		},
+		func(msg string) {
+			log.Trace().
+				Caller().
+				Str("machine_key", machineKey.ShortString()).
+				Str("node_key", registerRequest.NodeKey.ShortString()).
+				Str("node_key_old", registerRequest.OldNodeKey.ShortString()).
+				Str("node", registerRequest.Hostinfo.Hostname).
+				Str("followup", registerRequest.Followup).
+				Time("expiry", registerRequest.Expiry).
+				Msg(msg)
+		},
+		func(err error, msg string) {
+			log.Error().
+				Caller().
+				Str("machine_key", machineKey.ShortString()).
+				Str("node_key", registerRequest.NodeKey.ShortString()).
+				Str("node_key_old", registerRequest.OldNodeKey.ShortString()).
+				Str("node", registerRequest.Hostinfo.Hostname).
+				Str("followup", registerRequest.Followup).
+				Time("expiry", registerRequest.Expiry).
+				Err(err).
+				Msg(msg)
+		}
+}
+
+// handleRegister is the logic for registering a client.
 func (h *Headscale) handleRegister(
 	writer http.ResponseWriter,
 	req *http.Request,
 	registerRequest tailcfg.RegisterRequest,
 	machineKey key.MachinePublic,
-	isNoise bool,
 ) {
+	logInfo, logTrace, logErr := logAuthFunc(registerRequest, machineKey)
 	now := time.Now().UTC()
+	logTrace("handleRegister called, looking up machine in DB")
 	node, err := h.db.GetNodeByAnyKey(machineKey, registerRequest.NodeKey, registerRequest.OldNodeKey)
+	logTrace("handleRegister database lookup has returned")
 	if errors.Is(err, gorm.ErrRecordNotFound) {
 		// If the node has AuthKey set, handle registration via PreAuthKeys
 		if registerRequest.Auth.AuthKey != "" {
-			h.handleAuthKey(writer, registerRequest, machineKey, isNoise)
+			h.handleAuthKey(writer, registerRequest, machineKey)
 
 			return
 		}
@@ -45,49 +85,29 @@ func (h *Headscale) handleRegister(
 		// is that the client will hammer headscale with requests until it gets a
 		// successful RegisterResponse.
 		if registerRequest.Followup != "" {
-			if _, ok := h.registrationCache.Get(util.NodePublicKeyStripPrefix(registerRequest.NodeKey)); ok {
+			logTrace("register request is a followup")
-				log.Debug().
+			if _, ok := h.registrationCache.Get(machineKey.String()); ok {
-					Caller().
+				logTrace("Node is waiting for interactive login")
-					Str("node", registerRequest.Hostinfo.Hostname).
-					Str("machine_key", machineKey.ShortString()).
-					Str("node_key", registerRequest.NodeKey.ShortString()).
-					Str("node_key_old", registerRequest.OldNodeKey.ShortString()).
-					Str("follow_up", registerRequest.Followup).
-					Bool("noise", isNoise).
-					Msg("Node is waiting for interactive login")
 
 				select {
 				case <-req.Context().Done():
 					return
 				case <-time.After(registrationHoldoff):
-					h.handleNewNode(writer, registerRequest, machineKey, isNoise)
+					h.handleNewNode(writer, registerRequest, machineKey)
 
 					return
 				}
 			}
 		}
 
-		log.Info().
+		logInfo("Node not found in database, creating new")
-			Caller().
-			Str("node", registerRequest.Hostinfo.Hostname).
-			Str("machine_key", machineKey.ShortString()).
-			Str("node_key", registerRequest.NodeKey.ShortString()).
-			Str("node_key_old", registerRequest.OldNodeKey.ShortString()).
-			Str("follow_up", registerRequest.Followup).
-			Bool("noise", isNoise).
-			Msg("New node not yet in the database")
 
 		givenName, err := h.db.GenerateGivenName(
-			machineKey.String(),
+			machineKey,
 			registerRequest.Hostinfo.Hostname,
 		)
 		if err != nil {
-			log.Error().
+			logErr(err, "Failed to generate given name for node")
-				Caller().
-				Str("func", "RegistrationHandler").
-				Str("hostinfo.name", registerRequest.Hostinfo.Hostname).
-				Err(err).
-				Msg("Failed to generate given name for node")
 
 			return
 		}
@@ -97,31 +117,26 @@ func (h *Headscale) handleRegister(
 		// We create the node and then keep it around until a callback
 		// happens
 		newNode := types.Node{
-			MachineKey: util.MachinePublicKeyStripPrefix(machineKey),
+			MachineKey: machineKey,
 			Hostname:   registerRequest.Hostinfo.Hostname,
 			GivenName:  givenName,
-			NodeKey:    util.NodePublicKeyStripPrefix(registerRequest.NodeKey),
+			NodeKey:    registerRequest.NodeKey,
 			LastSeen:   &now,
 			Expiry:     &time.Time{},
 		}
 
 		if !registerRequest.Expiry.IsZero() {
-			log.Trace().
+			logTrace("Non-zero expiry time requested")
-				Caller().
-				Bool("noise", isNoise).
-				Str("node", registerRequest.Hostinfo.Hostname).
-				Time("expiry", registerRequest.Expiry).
-				Msg("Non-zero expiry time requested")
 			newNode.Expiry = &registerRequest.Expiry
 		}
 
 		h.registrationCache.Set(
-			newNode.NodeKey,
+			machineKey.String(),
 			newNode,
 			registerCacheExpiration,
 		)
 
-		h.handleNewNode(writer, registerRequest, machineKey, isNoise)
+		h.handleNewNode(writer, registerRequest, machineKey)
 
 		return
 	}
@@ -134,11 +149,7 @@ func (h *Headscale) handleRegister(
 		// (juan): For a while we had a bug where we were not storing the MachineKey for the nodes using the TS2021,
 		// due to a misunderstanding of the protocol https://github.com/juanfont/headscale/issues/1054
 		// So if we have a not valid MachineKey (but we were able to fetch the node with the NodeKeys), we update it.
-		var storedMachineKey key.MachinePublic
+		if err != nil || node.MachineKey.IsZero() {
-		err = storedMachineKey.UnmarshalText(
-			[]byte(util.MachinePublicKeyEnsurePrefix(node.MachineKey)),
-		)
-		if err != nil || storedMachineKey.IsZero() {
 			if err := h.db.NodeSetMachineKey(node, machineKey); err != nil {
 				log.Error().
 					Caller().
@@ -156,12 +167,12 @@ func (h *Headscale) handleRegister(
 		// - Trying to log out (sending a expiry in the past)
 		// - A valid, registered node, looking for /map
 		// - Expired node wanting to reauthenticate
-		if node.NodeKey == util.NodePublicKeyStripPrefix(registerRequest.NodeKey) {
+		if node.NodeKey.String() == registerRequest.NodeKey.String() {
 			// The client sends an Expiry in the past if the client is requesting to expire the key (aka logout)
 			//   https://github.com/tailscale/tailscale/blob/main/tailcfg/tailcfg.go#L648
 			if !registerRequest.Expiry.IsZero() &&
 				registerRequest.Expiry.UTC().Before(now) {
-				h.handleNodeLogOut(writer, *node, machineKey, isNoise)
+				h.handleNodeLogOut(writer, *node, machineKey)
 
 				return
 			}
@@ -169,21 +180,20 @@ func (h *Headscale) handleRegister(
 			// If node is not expired, and it is register, we have a already accepted this node,
 			// let it proceed with a valid registration
 			if !node.IsExpired() {
-				h.handleNodeWithValidRegistration(writer, *node, machineKey, isNoise)
+				h.handleNodeWithValidRegistration(writer, *node, machineKey)
 
 				return
 			}
 		}
 
 		// The NodeKey we have matches OldNodeKey, which means this is a refresh after a key expiration
-		if node.NodeKey == util.NodePublicKeyStripPrefix(registerRequest.OldNodeKey) &&
+		if node.NodeKey.String() == registerRequest.OldNodeKey.String() &&
 			!node.IsExpired() {
 			h.handleNodeKeyRefresh(
 				writer,
 				registerRequest,
 				*node,
 				machineKey,
-				isNoise,
 			)
 
 			return
@@ -198,7 +208,7 @@ func (h *Headscale) handleRegister(
 		}
 
 		// The node has expired or it is logged out
-		h.handleNodeExpiredOrLoggedOut(writer, registerRequest, *node, machineKey, isNoise)
+		h.handleNodeExpiredOrLoggedOut(writer, registerRequest, *node, machineKey)
 
 		// TODO(juan): RegisterRequest includes an Expiry time, that we could optionally use
 		node.Expiry = &time.Time{}
@@ -207,9 +217,9 @@ func (h *Headscale) handleRegister(
 		// we need to make sure the NodeKey matches the one in the request
 		// TODO(juan): What happens when using fast user switching between two
 		// headscale-managed tailnets?
-		node.NodeKey = util.NodePublicKeyStripPrefix(registerRequest.NodeKey)
+		node.NodeKey = registerRequest.NodeKey
 		h.registrationCache.Set(
-			util.NodePublicKeyStripPrefix(registerRequest.NodeKey),
+			machineKey.String(),
 			*node,
 			registerCacheExpiration,
 		)
@@ -219,7 +229,6 @@ func (h *Headscale) handleRegister(
 }
 
 // handleAuthKey contains the logic to manage auth key client registration
-// It is used both by the legacy and the new Noise protocol.
 // When using Noise, the machineKey is Zero.
 //
 // TODO: check if any locks are needed around IP allocation.
@@ -227,12 +236,10 @@ func (h *Headscale) handleAuthKey(
 	writer http.ResponseWriter,
 	registerRequest tailcfg.RegisterRequest,
 	machineKey key.MachinePublic,
-	isNoise bool,
 ) {
 	log.Debug().
 		Caller().
 		Str("node", registerRequest.Hostinfo.Hostname).
-		Bool("noise", isNoise).
 		Msgf("Processing auth key for %s", registerRequest.Hostinfo.Hostname)
 	resp := tailcfg.RegisterResponse{}
 
@@ -240,17 +247,15 @@ func (h *Headscale) handleAuthKey(
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Str("node", registerRequest.Hostinfo.Hostname).
 			Err(err).
 			Msg("Failed authentication via AuthKey")
 		resp.MachineAuthorized = false
 
-		respBody, err := mapper.MarshalResponse(resp, isNoise, h.privateKey2019, machineKey)
+		respBody, err := json.Marshal(resp)
 		if err != nil {
 			log.Error().
 				Caller().
-				Bool("noise", isNoise).
 				Str("node", registerRequest.Hostinfo.Hostname).
 				Err(err).
 				Msg("Cannot encode message")
@@ -267,14 +272,12 @@ func (h *Headscale) handleAuthKey(
 		if err != nil {
 			log.Error().
 				Caller().
-				Bool("noise", isNoise).
 				Err(err).
 				Msg("Failed to write response")
 		}
 
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Str("node", registerRequest.Hostinfo.Hostname).
 			Msg("Failed authentication via AuthKey")
 
@@ -290,11 +293,10 @@ func (h *Headscale) handleAuthKey(
 
 	log.Debug().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node", registerRequest.Hostinfo.Hostname).
 		Msg("Authentication key was valid, proceeding to acquire IP addresses")
 
-	nodeKey := util.NodePublicKeyStripPrefix(registerRequest.NodeKey)
+	nodeKey := registerRequest.NodeKey
 
 	// retrieve node information if it exist
 	// The error is not important, because if it does not
@@ -304,7 +306,6 @@ func (h *Headscale) handleAuthKey(
 	if node != nil {
 		log.Trace().
 			Caller().
-			Bool("noise", isNoise).
 			Str("node", node.Hostname).
 			Msg("node was already registered before, refreshing with new auth key")
 
@@ -314,7 +315,6 @@ func (h *Headscale) handleAuthKey(
 		if err != nil {
 			log.Error().
 				Caller().
-				Bool("noise", isNoise).
 				Str("node", node.Hostname).
 				Err(err).
 				Msg("Failed to refresh node")
@@ -322,7 +322,7 @@ func (h *Headscale) handleAuthKey(
 			return
 		}
 
-		aclTags := pak.Proto().AclTags
+		aclTags := pak.Proto().GetAclTags()
 		if len(aclTags) > 0 {
 			// This conditional preserves the existing behaviour, although SaaS would reset the tags on auth-key login
 			err = h.db.SetTags(node, aclTags)
@@ -330,7 +330,6 @@ func (h *Headscale) handleAuthKey(
 			if err != nil {
 				log.Error().
 					Caller().
-					Bool("noise", isNoise).
 					Str("node", node.Hostname).
 					Strs("aclTags", aclTags).
 					Err(err).
@@ -342,11 +341,10 @@ func (h *Headscale) handleAuthKey(
 	} else {
 		now := time.Now().UTC()
 
-		givenName, err := h.db.GenerateGivenName(util.MachinePublicKeyStripPrefix(machineKey), registerRequest.Hostinfo.Hostname)
+		givenName, err := h.db.GenerateGivenName(machineKey, registerRequest.Hostinfo.Hostname)
 		if err != nil {
 			log.Error().
 				Caller().
-				Bool("noise", isNoise).
 				Str("func", "RegistrationHandler").
 				Str("hostinfo.name", registerRequest.Hostinfo.Hostname).
 				Err(err).
@@ -359,13 +357,13 @@ func (h *Headscale) handleAuthKey(
 			Hostname:       registerRequest.Hostinfo.Hostname,
 			GivenName:      givenName,
 			UserID:         pak.User.ID,
-			MachineKey:     util.MachinePublicKeyStripPrefix(machineKey),
+			MachineKey:     machineKey,
 			RegisterMethod: util.RegisterMethodAuthKey,
 			Expiry:         &registerRequest.Expiry,
 			NodeKey:        nodeKey,
 			LastSeen:       &now,
 			AuthKeyID:      uint(pak.ID),
-			ForcedTags:     pak.Proto().AclTags,
+			ForcedTags:     pak.Proto().GetAclTags(),
 		}
 
 		node, err = h.db.RegisterNode(
@@ -374,7 +372,6 @@ func (h *Headscale) handleAuthKey(
 		if err != nil {
 			log.Error().
 				Caller().
-				Bool("noise", isNoise).
 				Err(err).
 				Msg("could not register node")
 			nodeRegistrations.WithLabelValues("new", util.RegisterMethodAuthKey, "error", pak.User.Name).
@@ -389,7 +386,6 @@ func (h *Headscale) handleAuthKey(
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Failed to use pre-auth key")
 		nodeRegistrations.WithLabelValues("new", util.RegisterMethodAuthKey, "error", pak.User.Name).
@@ -405,11 +401,10 @@ func (h *Headscale) handleAuthKey(
 	// Otherwise it will need to exec `tailscale up` twice to fetch the *LoginName*
 	resp.Login = *pak.User.TailscaleLogin()
 
-	respBody, err := mapper.MarshalResponse(resp, isNoise, h.privateKey2019, machineKey)
+	respBody, err := json.Marshal(resp)
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Str("node", registerRequest.Hostinfo.Hostname).
 			Err(err).
 			Msg("Cannot encode message")
@@ -427,54 +422,46 @@ func (h *Headscale) handleAuthKey(
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Failed to write response")
 	}
 
 	log.Info().
-		Bool("noise", isNoise).
 		Str("node", registerRequest.Hostinfo.Hostname).
 		Str("ips", strings.Join(node.IPAddresses.StringSlice(), ", ")).
 		Msg("Successfully authenticated via AuthKey")
 }
 
-// handleNewNode exposes for both legacy and Noise the functionality to get a URL
+// handleNewNode returns the authorisation URL to the client based on what type
-// for authorizing the node. This url is then showed to the user by the local Tailscale client.
+// of registration headscale is configured with.
+// This url is then showed to the user by the local Tailscale client.
 func (h *Headscale) handleNewNode(
 	writer http.ResponseWriter,
 	registerRequest tailcfg.RegisterRequest,
 	machineKey key.MachinePublic,
-	isNoise bool,
 ) {
+	logInfo, logTrace, logErr := logAuthFunc(registerRequest, machineKey)
+
 	resp := tailcfg.RegisterResponse{}
 
 	// The node registration is new, redirect the client to the registration URL
-	log.Debug().
+	logTrace("The node seems to be new, sending auth url")
-		Caller().
-		Bool("noise", isNoise).
-		Str("node", registerRequest.Hostinfo.Hostname).
-		Msg("The node seems to be new, sending auth url")
 
 	if h.oauth2Config != nil {
 		resp.AuthURL = fmt.Sprintf(
 			"%s/oidc/register/%s",
 			strings.TrimSuffix(h.cfg.ServerURL, "/"),
-			registerRequest.NodeKey,
+			machineKey.String(),
 		)
 	} else {
 		resp.AuthURL = fmt.Sprintf("%s/register/%s",
 			strings.TrimSuffix(h.cfg.ServerURL, "/"),
-			registerRequest.NodeKey)
+			machineKey.String())
 	}
 
-	respBody, err := mapper.MarshalResponse(resp, isNoise, h.privateKey2019, machineKey)
+	respBody, err := json.Marshal(resp)
 	if err != nil {
-		log.Error().
+		logErr(err, "Cannot encode message")
-			Caller().
-			Bool("noise", isNoise).
-			Err(err).
-			Msg("Cannot encode message")
 		http.Error(writer, "Internal server error", http.StatusInternalServerError)
 
 		return
@@ -484,31 +471,20 @@ func (h *Headscale) handleNewNode(
 	writer.WriteHeader(http.StatusOK)
 	_, err = writer.Write(respBody)
 	if err != nil {
-		log.Error().
+		logErr(err, "Failed to write response")
-			Bool("noise", isNoise).
-			Caller().
-			Err(err).
-			Msg("Failed to write response")
 	}
 
-	log.Info().
+	logInfo(fmt.Sprintf("Successfully sent auth url: %s", resp.AuthURL))
-		Caller().
-		Bool("noise", isNoise).
-		Str("AuthURL", resp.AuthURL).
-		Str("node", registerRequest.Hostinfo.Hostname).
-		Msg("Successfully sent auth url")
 }
 
 func (h *Headscale) handleNodeLogOut(
 	writer http.ResponseWriter,
 	node types.Node,
 	machineKey key.MachinePublic,
-	isNoise bool,
 ) {
 	resp := tailcfg.RegisterResponse{}
 
 	log.Info().
-		Bool("noise", isNoise).
 		Str("node", node.Hostname).
 		Msg("Client requested logout")
 
@@ -517,7 +493,6 @@ func (h *Headscale) handleNodeLogOut(
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Failed to expire node")
 		http.Error(writer, "Internal server error", http.StatusInternalServerError)
@@ -525,15 +500,27 @@ func (h *Headscale) handleNodeLogOut(
 		return
 	}
 
+	stateUpdate := types.StateUpdate{
+		Type: types.StatePeerChangedPatch,
+		ChangePatches: []*tailcfg.PeerChange{
+			{
+				NodeID:    tailcfg.NodeID(node.ID),
+				KeyExpiry: &now,
+			},
+		},
+	}
+	if stateUpdate.Valid() {
+		h.nodeNotifier.NotifyWithIgnore(stateUpdate, node.MachineKey.String())
+	}
+
 	resp.AuthURL = ""
 	resp.MachineAuthorized = false
 	resp.NodeKeyExpired = true
 	resp.User = *node.User.TailscaleUser()
-	respBody, err := mapper.MarshalResponse(resp, isNoise, h.privateKey2019, machineKey)
+	respBody, err := json.Marshal(resp)
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Cannot encode message")
 		http.Error(writer, "Internal server error", http.StatusInternalServerError)
@@ -546,7 +533,6 @@ func (h *Headscale) handleNodeLogOut(
 	_, err = writer.Write(respBody)
 	if err != nil {
 		log.Error().
-			Bool("noise", isNoise).
 			Caller().
 			Err(err).
 			Msg("Failed to write response")
@@ -568,7 +554,6 @@ func (h *Headscale) handleNodeLogOut(
 
 	log.Info().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node", node.Hostname).
 		Msg("Successfully logged out")
 }
@@ -577,14 +562,12 @@ func (h *Headscale) handleNodeWithValidRegistration(
 	writer http.ResponseWriter,
 	node types.Node,
 	machineKey key.MachinePublic,
-	isNoise bool,
 ) {
 	resp := tailcfg.RegisterResponse{}
 
 	// The node registration is valid, respond with redirect to /map
 	log.Debug().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node", node.Hostname).
 		Msg("Client is registered and we have the current NodeKey. All clear to /map")
 
@@ -593,11 +576,10 @@ func (h *Headscale) handleNodeWithValidRegistration(
 	resp.User = *node.User.TailscaleUser()
 	resp.Login = *node.User.TailscaleLogin()
 
-	respBody, err := mapper.MarshalResponse(resp, isNoise, h.privateKey2019, machineKey)
+	respBody, err := json.Marshal(resp)
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Cannot encode message")
 		nodeRegistrations.WithLabelValues("update", "web", "error", node.User.Name).
@@ -615,14 +597,12 @@ func (h *Headscale) handleNodeWithValidRegistration(
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Failed to write response")
 	}
 
 	log.Info().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node", node.Hostname).
 		Msg("Node successfully authorized")
 }
@@ -632,13 +612,11 @@ func (h *Headscale) handleNodeKeyRefresh(
 	registerRequest tailcfg.RegisterRequest,
 	node types.Node,
 	machineKey key.MachinePublic,
-	isNoise bool,
 ) {
 	resp := tailcfg.RegisterResponse{}
 
 	log.Info().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node", node.Hostname).
 		Msg("We have the OldNodeKey in the database. This is a key refresh")
 
@@ -655,11 +633,10 @@ func (h *Headscale) handleNodeKeyRefresh(
 
 	resp.AuthURL = ""
 	resp.User = *node.User.TailscaleUser()
-	respBody, err := mapper.MarshalResponse(resp, isNoise, h.privateKey2019, machineKey)
+	respBody, err := json.Marshal(resp)
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Cannot encode message")
 		http.Error(writer, "Internal server error", http.StatusInternalServerError)
@@ -673,14 +650,12 @@ func (h *Headscale) handleNodeKeyRefresh(
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Failed to write response")
 	}
 
 	log.Info().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node_key", registerRequest.NodeKey.ShortString()).
 		Str("old_node_key", registerRequest.OldNodeKey.ShortString()).
 		Str("node", node.Hostname).
@@ -692,12 +667,11 @@ func (h *Headscale) handleNodeExpiredOrLoggedOut(
 	registerRequest tailcfg.RegisterRequest,
 	node types.Node,
 	machineKey key.MachinePublic,
-	isNoise bool,
 ) {
 	resp := tailcfg.RegisterResponse{}
 
 	if registerRequest.Auth.AuthKey != "" {
-		h.handleAuthKey(writer, registerRequest, machineKey, isNoise)
+		h.handleAuthKey(writer, registerRequest, machineKey)
 
 		return
 	}
@@ -705,7 +679,6 @@ func (h *Headscale) handleNodeExpiredOrLoggedOut(
 	// The client has registered before, but has expired or logged out
 	log.Trace().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node", node.Hostname).
 		Str("machine_key", machineKey.ShortString()).
 		Str("node_key", registerRequest.NodeKey.ShortString()).
@@ -715,18 +688,17 @@ func (h *Headscale) handleNodeExpiredOrLoggedOut(
 	if h.oauth2Config != nil {
 		resp.AuthURL = fmt.Sprintf("%s/oidc/register/%s",
 			strings.TrimSuffix(h.cfg.ServerURL, "/"),
-			registerRequest.NodeKey)
+			machineKey.String())
 	} else {
 		resp.AuthURL = fmt.Sprintf("%s/register/%s",
 			strings.TrimSuffix(h.cfg.ServerURL, "/"),
-			registerRequest.NodeKey)
+			machineKey.String())
 	}
 
-	respBody, err := mapper.MarshalResponse(resp, isNoise, h.privateKey2019, machineKey)
+	respBody, err := json.Marshal(resp)
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Cannot encode message")
 		nodeRegistrations.WithLabelValues("reauth", "web", "error", node.User.Name).
@@ -744,14 +716,12 @@ func (h *Headscale) handleNodeExpiredOrLoggedOut(
 	if err != nil {
 		log.Error().
 			Caller().
-			Bool("noise", isNoise).
 			Err(err).
 			Msg("Failed to write response")
 	}
 
 	log.Trace().
 		Caller().
-		Bool("noise", isNoise).
 		Str("machine_key", machineKey.ShortString()).
 		Str("node_key", registerRequest.NodeKey.ShortString()).
 		Str("node_key_old", registerRequest.OldNodeKey.ShortString()).

hscontrol/auth_legacy.go

@@ -1,61 +0,0 @@
-//go:build ts2019
-
-package hscontrol
-
-import (
-	"io"
-	"net/http"
-
-	"github.com/gorilla/mux"
-	"github.com/juanfont/headscale/hscontrol/util"
-	"github.com/rs/zerolog/log"
-	"tailscale.com/tailcfg"
-	"tailscale.com/types/key"
-)
-
-// RegistrationHandler handles the actual registration process of a machine
-// Endpoint /machine/:mkey.
-func (h *Headscale) RegistrationHandler(
-	writer http.ResponseWriter,
-	req *http.Request,
-) {
-	vars := mux.Vars(req)
-	machineKeyStr, ok := vars["mkey"]
-	if !ok || machineKeyStr == "" {
-		log.Error().
-			Str("handler", "RegistrationHandler").
-			Msg("No machine ID in request")
-		http.Error(writer, "No machine ID in request", http.StatusBadRequest)
-
-		return
-	}
-
-	body, _ := io.ReadAll(req.Body)
-
-	var machineKey key.MachinePublic
-	err := machineKey.UnmarshalText([]byte(util.MachinePublicKeyEnsurePrefix(machineKeyStr)))
-	if err != nil {
-		log.Error().
-			Caller().
-			Err(err).
-			Msg("Cannot parse machine key")
-		nodeRegistrations.WithLabelValues("unknown", "web", "error", "unknown").Inc()
-		http.Error(writer, "Cannot parse machine key", http.StatusBadRequest)
-
-		return
-	}
-	registerRequest := tailcfg.RegisterRequest{}
-	err = util.DecodeAndUnmarshalNaCl(body, &registerRequest, &machineKey, h.privateKey2019)
-	if err != nil {
-		log.Error().
-			Caller().
-			Err(err).
-			Msg("Cannot decode message")
-		nodeRegistrations.WithLabelValues("unknown", "web", "error", "unknown").Inc()
-		http.Error(writer, "Cannot decode message", http.StatusBadRequest)
-
-		return
-	}
-
-	h.handleRegister(writer, req, registerRequest, machineKey, false)
-}

hscontrol/auth_noise.go

@@ -39,7 +39,19 @@ func (ns *noiseServer) NoiseRegistrationHandler(
 		return
 	}
 
+	// Reject unsupported versions
+	if registerRequest.Version < MinimumCapVersion {
+		log.Info().
+			Caller().
+			Int("min_version", int(MinimumCapVersion)).
+			Int("client_version", int(registerRequest.Version)).
+			Msg("unsupported client connected")
+		http.Error(writer, "Internal error", http.StatusBadRequest)
+
+		return
+	}
+
 	ns.nodeKey = registerRequest.NodeKey
 
-	ns.headscale.handleRegister(writer, req, registerRequest, ns.conn.Peer(), true)
+	ns.headscale.handleRegister(writer, req, registerRequest, ns.conn.Peer())
 }

hscontrol/db/addresses_test.go

@@ -35,9 +35,6 @@ func (s *Suite) TestGetUsedIps(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -83,9 +80,6 @@ func (s *Suite) TestGetMultiIp(c *check.C) {
 
 		node := types.Node{
 			ID:             uint64(index),
-			MachineKey:     "foo",
-			NodeKey:        "bar",
-			DiscoKey:       "faa",
 			Hostname:       "testnode",
 			UserID:         user.ID,
 			RegisterMethod: util.RegisterMethodAuthKey,
@@ -173,9 +167,6 @@ func (s *Suite) TestGetAvailableIpNodeWithoutIP(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,

hscontrol/db/db.go

@@ -2,13 +2,16 @@ package db
 
 import (
 	"context"
+	"database/sql"
 	"errors"
 	"fmt"
 	"net/netip"
+	"strings"
 	"sync"
 	"time"
 
 	"github.com/glebarez/sqlite"
+	"github.com/go-gormigrate/gormigrate/v2"
 	"github.com/juanfont/headscale/hscontrol/notifier"
 	"github.com/juanfont/headscale/hscontrol/types"
 	"github.com/juanfont/headscale/hscontrol/util"
@@ -19,15 +22,11 @@ import (
 )
 
 const (
-	dbVersion = "1"
+	Postgres = "postgres"
-	Postgres  = "postgres"
+	Sqlite   = "sqlite3"
-	Sqlite    = "sqlite3"
 )
 
-var (
+var errDatabaseNotSupported = errors.New("database type not supported")
-	errValueNotFound        = errors.New("not found")
-	errDatabaseNotSupported = errors.New("database type not supported")
-)
 
 // KV is a key-value store in a psql table. For future use...
 // TODO(kradalby): Is this used for anything?
@@ -62,6 +61,261 @@ func NewHeadscaleDatabase(
 		return nil, err
 	}
 
+	migrations := gormigrate.New(dbConn, gormigrate.DefaultOptions, []*gormigrate.Migration{
+		// New migrations should be added as transactions at the end of this list.
+		// The initial commit here is quite messy, completely out of order and
+		// has no versioning and is the tech debt of not having versioned migrations
+		// prior to this point. This first migration is all DB changes to bring a DB
+		// up to 0.23.0.
+		{
+			ID: "202312101416",
+			Migrate: func(tx *gorm.DB) error {
+				if dbType == Postgres {
+					tx.Exec(`create extension if not exists "uuid-ossp";`)
+				}
+
+				_ = tx.Migrator().RenameTable("namespaces", "users")
+
+				// the big rename from Machine to Node
+				_ = tx.Migrator().RenameTable("machines", "nodes")
+				_ = tx.Migrator().RenameColumn(&types.Route{}, "machine_id", "node_id")
+
+				err = tx.AutoMigrate(types.User{})
+				if err != nil {
+					return err
+				}
+
+				_ = tx.Migrator().RenameColumn(&types.Node{}, "namespace_id", "user_id")
+				_ = tx.Migrator().RenameColumn(&types.PreAuthKey{}, "namespace_id", "user_id")
+
+				_ = tx.Migrator().RenameColumn(&types.Node{}, "ip_address", "ip_addresses")
+				_ = tx.Migrator().RenameColumn(&types.Node{}, "name", "hostname")
+
+				// GivenName is used as the primary source of DNS names, make sure
+				// the field is populated and normalized if it was not when the
+				// node was registered.
+				_ = tx.Migrator().RenameColumn(&types.Node{}, "nickname", "given_name")
+
+				// If the Node table has a column for registered,
+				// find all occourences of "false" and drop them. Then
+				// remove the column.
+				if tx.Migrator().HasColumn(&types.Node{}, "registered") {
+					log.Info().
+						Msg(`Database has legacy "registered" column in node, removing...`)
+
+					nodes := types.Nodes{}
+					if err := tx.Not("registered").Find(&nodes).Error; err != nil {
+						log.Error().Err(err).Msg("Error accessing db")
+					}
+
+					for _, node := range nodes {
+						log.Info().
+							Str("node", node.Hostname).
+							Str("machine_key", node.MachineKey.ShortString()).
+							Msg("Deleting unregistered node")
+						if err := tx.Delete(&types.Node{}, node.ID).Error; err != nil {
+							log.Error().
+								Err(err).
+								Str("node", node.Hostname).
+								Str("machine_key", node.MachineKey.ShortString()).
+								Msg("Error deleting unregistered node")
+						}
+					}
+
+					err := tx.Migrator().DropColumn(&types.Node{}, "registered")
+					if err != nil {
+						log.Error().Err(err).Msg("Error dropping registered column")
+					}
+				}
+
+				err = tx.AutoMigrate(&types.Route{})
+				if err != nil {
+					return err
+				}
+
+				err = tx.AutoMigrate(&types.Node{})
+				if err != nil {
+					return err
+				}
+
+				// Ensure all keys have correct prefixes
+				// https://github.com/tailscale/tailscale/blob/main/types/key/node.go#L35
+				type result struct {
+					ID         uint64
+					MachineKey string
+					NodeKey    string
+					DiscoKey   string
+				}
+				var results []result
+				err = tx.Raw("SELECT id, node_key, machine_key, disco_key FROM nodes").Find(&results).Error
+				if err != nil {
+					return err
+				}
+
+				for _, node := range results {
+					mKey := node.MachineKey
+					if !strings.HasPrefix(node.MachineKey, "mkey:") {
+						mKey = "mkey:" + node.MachineKey
+					}
+					nKey := node.NodeKey
+					if !strings.HasPrefix(node.NodeKey, "nodekey:") {
+						nKey = "nodekey:" + node.NodeKey
+					}
+
+					dKey := node.DiscoKey
+					if !strings.HasPrefix(node.DiscoKey, "discokey:") {
+						dKey = "discokey:" + node.DiscoKey
+					}
+
+					err := tx.Exec(
+						"UPDATE nodes SET machine_key = @mKey, node_key = @nKey, disco_key = @dKey WHERE ID = @id",
+						sql.Named("mKey", mKey),
+						sql.Named("nKey", nKey),
+						sql.Named("dKey", dKey),
+						sql.Named("id", node.ID),
+					).Error
+					if err != nil {
+						return err
+					}
+				}
+
+				if tx.Migrator().HasColumn(&types.Node{}, "enabled_routes") {
+					log.Info().Msgf("Database has legacy enabled_routes column in node, migrating...")
+
+					type NodeAux struct {
+						ID            uint64
+						EnabledRoutes types.IPPrefixes
+					}
+
+					nodesAux := []NodeAux{}
+					err := tx.Table("nodes").Select("id, enabled_routes").Scan(&nodesAux).Error
+					if err != nil {
+						log.Fatal().Err(err).Msg("Error accessing db")
+					}
+					for _, node := range nodesAux {
+						for _, prefix := range node.EnabledRoutes {
+							if err != nil {
+								log.Error().
+									Err(err).
+									Str("enabled_route", prefix.String()).
+									Msg("Error parsing enabled_route")
+
+								continue
+							}
+
+							err = tx.Preload("Node").
+								Where("node_id = ? AND prefix = ?", node.ID, types.IPPrefix(prefix)).
+								First(&types.Route{}).
+								Error
+							if err == nil {
+								log.Info().
+									Str("enabled_route", prefix.String()).
+									Msg("Route already migrated to new table, skipping")
+
+								continue
+							}
+
+							route := types.Route{
+								NodeID:     node.ID,
+								Advertised: true,
+								Enabled:    true,
+								Prefix:     types.IPPrefix(prefix),
+							}
+							if err := tx.Create(&route).Error; err != nil {
+								log.Error().Err(err).Msg("Error creating route")
+							} else {
+								log.Info().
+									Uint64("node_id", route.NodeID).
+									Str("prefix", prefix.String()).
+									Msg("Route migrated")
+							}
+						}
+					}
+
+					err = tx.Migrator().DropColumn(&types.Node{}, "enabled_routes")
+					if err != nil {
+						log.Error().Err(err).Msg("Error dropping enabled_routes column")
+					}
+				}
+
+				if tx.Migrator().HasColumn(&types.Node{}, "given_name") {
+					nodes := types.Nodes{}
+					if err := tx.Find(&nodes).Error; err != nil {
+						log.Error().Err(err).Msg("Error accessing db")
+					}
+
+					for item, node := range nodes {
+						if node.GivenName == "" {
+							normalizedHostname, err := util.NormalizeToFQDNRulesConfigFromViper(
+								node.Hostname,
+							)
+							if err != nil {
+								log.Error().
+									Caller().
+									Str("hostname", node.Hostname).
+									Err(err).
+									Msg("Failed to normalize node hostname in DB migration")
+							}
+
+							err = tx.Model(nodes[item]).Updates(types.Node{
+								GivenName: normalizedHostname,
+							}).Error
+							if err != nil {
+								log.Error().
+									Caller().
+									Str("hostname", node.Hostname).
+									Err(err).
+									Msg("Failed to save normalized node name in DB migration")
+							}
+						}
+					}
+				}
+
+				err = tx.AutoMigrate(&KV{})
+				if err != nil {
+					return err
+				}
+
+				err = tx.AutoMigrate(&types.PreAuthKey{})
+				if err != nil {
+					return err
+				}
+
+				err = tx.AutoMigrate(&types.PreAuthKeyACLTag{})
+				if err != nil {
+					return err
+				}
+
+				_ = tx.Migrator().DropTable("shared_machines")
+
+				err = tx.AutoMigrate(&types.APIKey{})
+				if err != nil {
+					return err
+				}
+
+				return nil
+			},
+			Rollback: func(tx *gorm.DB) error {
+				return nil
+			},
+		},
+		{
+			// drop key-value table, it is not used, and has not contained
+			// useful data for a long time or ever.
+			ID: "202312101430",
+			Migrate: func(tx *gorm.DB) error {
+				return tx.Migrator().DropTable("kvs")
+			},
+			Rollback: func(tx *gorm.DB) error {
+				return nil
+			},
+		},
+	})
+
+	if err = migrations.Migrate(); err != nil {
+		log.Fatal().Err(err).Msgf("Migration failed: %v", err)
+	}
+
 	db := HSDatabase{
 		db:       dbConn,
 		notifier: notifier,
@@ -70,191 +324,6 @@ func NewHeadscaleDatabase(
 		baseDomain: baseDomain,
 	}
 
-	log.Debug().Msgf("database %#v", dbConn)
-
-	if dbType == Postgres {
-		dbConn.Exec(`create extension if not exists "uuid-ossp";`)
-	}
-
-	_ = dbConn.Migrator().RenameTable("namespaces", "users")
-
-	// the big rename from Machine to Node
-	_ = dbConn.Migrator().RenameTable("machines", "nodes")
-	_ = dbConn.Migrator().RenameColumn(&types.Route{}, "machine_id", "node_id")
-
-	err = dbConn.AutoMigrate(types.User{})
-	if err != nil {
-		return nil, err
-	}
-
-	_ = dbConn.Migrator().RenameColumn(&types.Node{}, "namespace_id", "user_id")
-	_ = dbConn.Migrator().RenameColumn(&types.PreAuthKey{}, "namespace_id", "user_id")
-
-	_ = dbConn.Migrator().RenameColumn(&types.Node{}, "ip_address", "ip_addresses")
-	_ = dbConn.Migrator().RenameColumn(&types.Node{}, "name", "hostname")
-
-	// GivenName is used as the primary source of DNS names, make sure
-	// the field is populated and normalized if it was not when the
-	// node was registered.
-	_ = dbConn.Migrator().RenameColumn(&types.Node{}, "nickname", "given_name")
-
-	// If the MacNodehine table has a column for registered,
-	// find all occourences of "false" and drop them. Then
-	// remove the column.
-	if dbConn.Migrator().HasColumn(&types.Node{}, "registered") {
-		log.Info().
-			Msg(`Database has legacy "registered" column in node, removing...`)
-
-		nodes := types.Nodes{}
-		if err := dbConn.Not("registered").Find(&nodes).Error; err != nil {
-			log.Error().Err(err).Msg("Error accessing db")
-		}
-
-		for _, node := range nodes {
-			log.Info().
-				Str("node", node.Hostname).
-				Str("machine_key", node.MachineKey).
-				Msg("Deleting unregistered node")
-			if err := dbConn.Delete(&types.Node{}, node.ID).Error; err != nil {
-				log.Error().
-					Err(err).
-					Str("node", node.Hostname).
-					Str("machine_key", node.MachineKey).
-					Msg("Error deleting unregistered node")
-			}
-		}
-
-		err := dbConn.Migrator().DropColumn(&types.Node{}, "registered")
-		if err != nil {
-			log.Error().Err(err).Msg("Error dropping registered column")
-		}
-	}
-
-	err = dbConn.AutoMigrate(&types.Route{})
-	if err != nil {
-		return nil, err
-	}
-
-	if dbConn.Migrator().HasColumn(&types.Node{}, "enabled_routes") {
-		log.Info().Msgf("Database has legacy enabled_routes column in node, migrating...")
-
-		type NodeAux struct {
-			ID            uint64
-			EnabledRoutes types.IPPrefixes
-		}
-
-		nodesAux := []NodeAux{}
-		err := dbConn.Table("nodes").Select("id, enabled_routes").Scan(&nodesAux).Error
-		if err != nil {
-			log.Fatal().Err(err).Msg("Error accessing db")
-		}
-		for _, node := range nodesAux {
-			for _, prefix := range node.EnabledRoutes {
-				if err != nil {
-					log.Error().
-						Err(err).
-						Str("enabled_route", prefix.String()).
-						Msg("Error parsing enabled_route")
-
-					continue
-				}
-
-				err = dbConn.Preload("Node").
-					Where("node_id = ? AND prefix = ?", node.ID, types.IPPrefix(prefix)).
-					First(&types.Route{}).
-					Error
-				if err == nil {
-					log.Info().
-						Str("enabled_route", prefix.String()).
-						Msg("Route already migrated to new table, skipping")
-
-					continue
-				}
-
-				route := types.Route{
-					NodeID:     node.ID,
-					Advertised: true,
-					Enabled:    true,
-					Prefix:     types.IPPrefix(prefix),
-				}
-				if err := dbConn.Create(&route).Error; err != nil {
-					log.Error().Err(err).Msg("Error creating route")
-				} else {
-					log.Info().
-						Uint64("node_id", route.NodeID).
-						Str("prefix", prefix.String()).
-						Msg("Route migrated")
-				}
-			}
-		}
-
-		err = dbConn.Migrator().DropColumn(&types.Node{}, "enabled_routes")
-		if err != nil {
-			log.Error().Err(err).Msg("Error dropping enabled_routes column")
-		}
-	}
-
-	err = dbConn.AutoMigrate(&types.Node{})
-	if err != nil {
-		return nil, err
-	}
-
-	if dbConn.Migrator().HasColumn(&types.Node{}, "given_name") {
-		nodes := types.Nodes{}
-		if err := dbConn.Find(&nodes).Error; err != nil {
-			log.Error().Err(err).Msg("Error accessing db")
-		}
-
-		for item, node := range nodes {
-			if node.GivenName == "" {
-				normalizedHostname, err := util.NormalizeToFQDNRulesConfigFromViper(
-					node.Hostname,
-				)
-				if err != nil {
-					log.Error().
-						Caller().
-						Str("hostname", node.Hostname).
-						Err(err).
-						Msg("Failed to normalize node hostname in DB migration")
-				}
-
-				err = db.RenameNode(nodes[item], normalizedHostname)
-				if err != nil {
-					log.Error().
-						Caller().
-						Str("hostname", node.Hostname).
-						Err(err).
-						Msg("Failed to save normalized node name in DB migration")
-				}
-			}
-		}
-	}
-
-	err = dbConn.AutoMigrate(&KV{})
-	if err != nil {
-		return nil, err
-	}
-
-	err = dbConn.AutoMigrate(&types.PreAuthKey{})
-	if err != nil {
-		return nil, err
-	}
-
-	err = dbConn.AutoMigrate(&types.PreAuthKeyACLTag{})
-	if err != nil {
-		return nil, err
-	}
-
-	_ = dbConn.Migrator().DropTable("shared_machines")
-
-	err = dbConn.AutoMigrate(&types.APIKey{})
-	if err != nil {
-		return nil, err
-	}
-
-	// TODO(kradalby): is this needed?
-	err = db.setValue("db_version", dbVersion)
-
 	return &db, err
 }
 
@@ -304,39 +373,6 @@ func openDB(dbType, connectionAddr string, debug bool) (*gorm.DB, error) {
 	)
 }
 
-// getValue returns the value for the given key in KV.
-func (hsdb *HSDatabase) getValue(key string) (string, error) {
-	var row KV
-	if result := hsdb.db.First(&row, "key = ?", key); errors.Is(
-		result.Error,
-		gorm.ErrRecordNotFound,
-	) {
-		return "", errValueNotFound
-	}
-
-	return row.Value, nil
-}
-
-// setValue sets value for the given key in KV.
-func (hsdb *HSDatabase) setValue(key string, value string) error {
-	keyValue := KV{
-		Key:   key,
-		Value: value,
-	}
-
-	if _, err := hsdb.getValue(key); err == nil {
-		hsdb.db.Model(&keyValue).Where("key = ?", key).Update("value", value)
-
-		return nil
-	}
-
-	if err := hsdb.db.Create(keyValue).Error; err != nil {
-		return fmt.Errorf("failed to create key value pair in the database: %w", err)
-	}
-
-	return nil
-}
-
 func (hsdb *HSDatabase) PingDB(ctx context.Context) error {
 	ctx, cancel := context.WithTimeout(ctx, time.Second)
 	defer cancel()

hscontrol/db/node.go

@@ -55,17 +55,12 @@ func (hsdb *HSDatabase) listPeers(node *types.Node) (types.Nodes, error) {
 		Preload("User").
 		Preload("Routes").
 		Where("node_key <> ?",
-			node.NodeKey).Find(&nodes).Error; err != nil {
+			node.NodeKey.String()).Find(&nodes).Error; err != nil {
 		return types.Nodes{}, err
 	}
 
 	sort.Slice(nodes, func(i, j int) bool { return nodes[i].ID < nodes[j].ID })
 
-	log.Trace().
-		Caller().
-		Str("node", node.Hostname).
-		Msgf("Found peers: %s", nodes.String())
-
 	return nodes, nil
 }
 
@@ -176,13 +171,19 @@ func (hsdb *HSDatabase) GetNodeByMachineKey(
 	hsdb.mu.RLock()
 	defer hsdb.mu.RUnlock()
 
+	return hsdb.getNodeByMachineKey(machineKey)
+}
+
+func (hsdb *HSDatabase) getNodeByMachineKey(
+	machineKey key.MachinePublic,
+) (*types.Node, error) {
 	mach := types.Node{}
 	if result := hsdb.db.
 		Preload("AuthKey").
 		Preload("AuthKey.User").
 		Preload("User").
 		Preload("Routes").
-		First(&mach, "machine_key = ?", util.MachinePublicKeyStripPrefix(machineKey)); result.Error != nil {
+		First(&mach, "machine_key = ?", machineKey.String()); result.Error != nil {
 		return nil, result.Error
 	}
 
@@ -203,7 +204,7 @@ func (hsdb *HSDatabase) GetNodeByNodeKey(
 		Preload("User").
 		Preload("Routes").
 		First(&node, "node_key = ?",
-			util.NodePublicKeyStripPrefix(nodeKey)); result.Error != nil {
+			nodeKey.String()); result.Error != nil {
 		return nil, result.Error
 	}
 
@@ -224,9 +225,9 @@ func (hsdb *HSDatabase) GetNodeByAnyKey(
 		Preload("User").
 		Preload("Routes").
 		First(&node, "machine_key = ? OR node_key = ? OR node_key = ?",
-			util.MachinePublicKeyStripPrefix(machineKey),
+			machineKey.String(),
-			util.NodePublicKeyStripPrefix(nodeKey),
+			nodeKey.String(),
-			util.NodePublicKeyStripPrefix(oldNodeKey)); result.Error != nil {
+			oldNodeKey.String()); result.Error != nil {
 		return nil, result.Error
 	}
 
@@ -252,6 +253,10 @@ func (hsdb *HSDatabase) SetTags(
 	hsdb.mu.Lock()
 	defer hsdb.mu.Unlock()
 
+	if len(tags) == 0 {
+		return nil
+	}
+
 	newTags := []string{}
 	for _, tag := range tags {
 		if !util.StringOrPrefixListContains(newTags, tag) {
@@ -265,10 +270,14 @@ func (hsdb *HSDatabase) SetTags(
 		return fmt.Errorf("failed to update tags for node in the database: %w", err)
 	}
 
-	hsdb.notifier.NotifyWithIgnore(types.StateUpdate{
+	stateUpdate := types.StateUpdate{
-		Type:    types.StatePeerChanged,
+		Type:        types.StatePeerChanged,
-		Changed: types.Nodes{node},
+		ChangeNodes: types.Nodes{node},
-	}, node.MachineKey)
+		Message:     "called from db.SetTags",
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyWithIgnore(stateUpdate, node.MachineKey.String())
+	}
 
 	return nil
 }
@@ -301,10 +310,14 @@ func (hsdb *HSDatabase) RenameNode(node *types.Node, newName string) error {
 		return fmt.Errorf("failed to rename node in the database: %w", err)
 	}
 
-	hsdb.notifier.NotifyWithIgnore(types.StateUpdate{
+	stateUpdate := types.StateUpdate{
-		Type:    types.StatePeerChanged,
+		Type:        types.StatePeerChanged,
-		Changed: types.Nodes{node},
+		ChangeNodes: types.Nodes{node},
-	}, node.MachineKey)
+		Message:     "called from db.RenameNode",
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyWithIgnore(stateUpdate, node.MachineKey.String())
+	}
 
 	return nil
 }
@@ -327,10 +340,28 @@ func (hsdb *HSDatabase) nodeSetExpiry(node *types.Node, expiry time.Time) error
 		)
 	}
 
-	hsdb.notifier.NotifyWithIgnore(types.StateUpdate{
+	node.Expiry = &expiry
-		Type:    types.StatePeerChanged,
+
-		Changed: types.Nodes{node},
+	stateSelfUpdate := types.StateUpdate{
-	}, node.MachineKey)
+		Type:        types.StateSelfUpdate,
+		ChangeNodes: types.Nodes{node},
+	}
+	if stateSelfUpdate.Valid() {
+		hsdb.notifier.NotifyByMachineKey(stateSelfUpdate, node.MachineKey)
+	}
+
+	stateUpdate := types.StateUpdate{
+		Type: types.StatePeerChangedPatch,
+		ChangePatches: []*tailcfg.PeerChange{
+			{
+				NodeID:    tailcfg.NodeID(node.ID),
+				KeyExpiry: &expiry,
+			},
+		},
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyWithIgnore(stateUpdate, node.MachineKey.String())
+	}
 
 	return nil
 }
@@ -354,10 +385,13 @@ func (hsdb *HSDatabase) deleteNode(node *types.Node) error {
 		return err
 	}
 
-	hsdb.notifier.NotifyAll(types.StateUpdate{
+	stateUpdate := types.StateUpdate{
 		Type:    types.StatePeerRemoved,
 		Removed: []tailcfg.NodeID{tailcfg.NodeID(node.ID)},
-	})
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyAll(stateUpdate)
+	}
 
 	return nil
 }
@@ -376,7 +410,7 @@ func (hsdb *HSDatabase) UpdateLastSeen(node *types.Node) error {
 
 func (hsdb *HSDatabase) RegisterNodeFromAuthCallback(
 	cache *cache.Cache,
-	nodeKeyStr string,
+	mkey key.MachinePublic,
 	userName string,
 	nodeExpiry *time.Time,
 	registrationMethod string,
@@ -384,20 +418,14 @@ func (hsdb *HSDatabase) RegisterNodeFromAuthCallback(
 	hsdb.mu.Lock()
 	defer hsdb.mu.Unlock()
 
-	nodeKey := key.NodePublic{}
-	err := nodeKey.UnmarshalText([]byte(nodeKeyStr))
-	if err != nil {
-		return nil, err
-	}
-
 	log.Debug().
-		Str("nodeKey", nodeKey.ShortString()).
+		Str("machine_key", mkey.ShortString()).
 		Str("userName", userName).
 		Str("registrationMethod", registrationMethod).
 		Str("expiresAt", fmt.Sprintf("%v", nodeExpiry)).
 		Msg("Registering node from API/CLI or auth callback")
 
-	if nodeInterface, ok := cache.Get(util.NodePublicKeyStripPrefix(nodeKey)); ok {
+	if nodeInterface, ok := cache.Get(mkey.String()); ok {
 		if registrationNode, ok := nodeInterface.(types.Node); ok {
 			user, err := hsdb.getUser(userName)
 			if err != nil {
@@ -425,7 +453,7 @@ func (hsdb *HSDatabase) RegisterNodeFromAuthCallback(
 			)
 
 			if err == nil {
-				cache.Delete(nodeKeyStr)
+				cache.Delete(mkey.String())
 			}
 
 			return node, err
@@ -448,8 +476,8 @@ func (hsdb *HSDatabase) RegisterNode(node types.Node) (*types.Node, error) {
 func (hsdb *HSDatabase) registerNode(node types.Node) (*types.Node, error) {
 	log.Debug().
 		Str("node", node.Hostname).
-		Str("machine_key", node.MachineKey).
+		Str("machine_key", node.MachineKey.ShortString()).
-		Str("node_key", node.NodeKey).
+		Str("node_key", node.NodeKey.ShortString()).
 		Str("user", node.User.Name).
 		Msg("Registering node")
 
@@ -464,8 +492,8 @@ func (hsdb *HSDatabase) registerNode(node types.Node) (*types.Node, error) {
 		log.Trace().
 			Caller().
 			Str("node", node.Hostname).
-			Str("machine_key", node.MachineKey).
+			Str("machine_key", node.MachineKey.ShortString()).
-			Str("node_key", node.NodeKey).
+			Str("node_key", node.NodeKey.ShortString()).
 			Str("user", node.User.Name).
 			Msg("Node authorized again")
 
@@ -507,7 +535,7 @@ func (hsdb *HSDatabase) NodeSetNodeKey(node *types.Node, nodeKey key.NodePublic)
 	defer hsdb.mu.Unlock()
 
 	if err := hsdb.db.Model(node).Updates(types.Node{
-		NodeKey: util.NodePublicKeyStripPrefix(nodeKey),
+		NodeKey: nodeKey,
 	}).Error; err != nil {
 		return err
 	}
@@ -524,7 +552,7 @@ func (hsdb *HSDatabase) NodeSetMachineKey(
 	defer hsdb.mu.Unlock()
 
 	if err := hsdb.db.Model(node).Updates(types.Node{
-		MachineKey: util.MachinePublicKeyStripPrefix(machineKey),
+		MachineKey: machineKey,
 	}).Error; err != nil {
 		return err
 	}
@@ -635,20 +663,6 @@ func (hsdb *HSDatabase) IsRoutesEnabled(node *types.Node, routeStr string) bool
 	return false
 }
 
-func (hsdb *HSDatabase) ListOnlineNodes(
-	node *types.Node,
-) (map[tailcfg.NodeID]bool, error) {
-	hsdb.mu.RLock()
-	defer hsdb.mu.RUnlock()
-
-	peers, err := hsdb.listPeers(node)
-	if err != nil {
-		return nil, err
-	}
-
-	return peers.OnlineNodeMap(), nil
-}
-
 // enableRoutes enables new routes based on a list of new routes.
 func (hsdb *HSDatabase) enableRoutes(node *types.Node, routeStrs ...string) error {
 	newRoutes := make([]netip.Prefix, len(routeStrs))
@@ -700,10 +714,30 @@ func (hsdb *HSDatabase) enableRoutes(node *types.Node, routeStrs ...string) erro
 		}
 	}
 
-	hsdb.notifier.NotifyWithIgnore(types.StateUpdate{
+	// Ensure the node has the latest routes when notifying the other
-		Type:    types.StatePeerChanged,
+	// nodes
-		Changed: types.Nodes{node},
+	nRoutes, err := hsdb.getNodeRoutes(node)
-	}, node.MachineKey)
+	if err != nil {
+		return fmt.Errorf("failed to read back routes: %w", err)
+	}
+
+	node.Routes = nRoutes
+
+	log.Trace().
+		Caller().
+		Str("node", node.Hostname).
+		Strs("routes", routeStrs).
+		Msg("enabling routes")
+
+	stateUpdate := types.StateUpdate{
+		Type:        types.StatePeerChanged,
+		ChangeNodes: types.Nodes{node},
+		Message:     "called from db.enableRoutes",
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyWithIgnore(
+			stateUpdate, node.MachineKey.String())
+	}
 
 	return nil
 }
@@ -734,7 +768,10 @@ func generateGivenName(suppliedName string, randomSuffix bool) (string, error) {
 	return normalizedHostname, nil
 }
 
-func (hsdb *HSDatabase) GenerateGivenName(machineKey string, suppliedName string) (string, error) {
+func (hsdb *HSDatabase) GenerateGivenName(
+	mkey key.MachinePublic,
+	suppliedName string,
+) (string, error) {
 	hsdb.mu.RLock()
 	defer hsdb.mu.RUnlock()
 
@@ -749,17 +786,22 @@ func (hsdb *HSDatabase) GenerateGivenName(machineKey string, suppliedName string
 		return "", err
 	}
 
-	for _, node := range nodes {
+	var nodeFound *types.Node
-		if node.MachineKey != machineKey && node.GivenName == givenName {
+	for idx, node := range nodes {
-			postfixedName, err := generateGivenName(suppliedName, true)
+		if node.GivenName == givenName {
-			if err != nil {
+			nodeFound = nodes[idx]
-				return "", err
-			}
-
-			givenName = postfixedName
 		}
 	}
 
+	if nodeFound != nil && nodeFound.MachineKey.String() != mkey.String() {
+		postfixedName, err := generateGivenName(suppliedName, true)
+		if err != nil {
+			return "", err
+		}
+
+		givenName = postfixedName
+	}
+
 	return givenName, nil
 }
 
@@ -824,52 +866,69 @@ func (hsdb *HSDatabase) ExpireExpiredNodes(lastCheck time.Time) time.Time {
 	// checked everything.
 	started := time.Now()
 
-	users, err := hsdb.listUsers()
+	expiredNodes := make([]*types.Node, 0)
+
+	nodes, err := hsdb.listNodes()
 	if err != nil {
-		log.Error().Err(err).Msg("Error listing users")
+		log.Error().
+			Err(err).
+			Msg("Error listing nodes to find expired nodes")
 
 		return time.Unix(0, 0)
 	}
+	for index, node := range nodes {
+		if node.IsExpired() &&
+			// TODO(kradalby): Replace this, it is very spammy
+			// It will notify about all nodes that has been expired.
+			// It should only notify about expired nodes since _last check_.
+			node.Expiry.After(lastCheck) {
+			expiredNodes = append(expiredNodes, &nodes[index])
 
-	for _, user := range users {
+			// Do not use setNodeExpiry as that has a notifier hook, which
-		nodes, err := hsdb.listNodesByUser(user.Name)
+			// can cause a deadlock, we are updating all changed nodes later
-		if err != nil {
+			// and there is no point in notifiying twice.
-			log.Error().
+			if err := hsdb.db.Model(nodes[index]).Updates(types.Node{
-				Err(err).
+				Expiry: &started,
-				Str("user", user.Name).
+			}).Error; err != nil {
-				Msg("Error listing nodes in user")
+				log.Error().
-
+					Err(err).
-			return time.Unix(0, 0)
+					Str("node", node.Hostname).
-		}
+					Str("name", node.GivenName).
-
+					Msg("🤮 Cannot expire node")
-		expired := make([]tailcfg.NodeID, 0)
+			} else {
-		for index, node := range nodes {
+				log.Info().
-			if node.IsExpired() &&
+					Str("node", node.Hostname).
-				node.Expiry.After(lastCheck) {
+					Str("name", node.GivenName).
-				expired = append(expired, tailcfg.NodeID(node.ID))
+					Msg("Node successfully expired")
-
-				now := time.Now()
-				err := hsdb.nodeSetExpiry(nodes[index], now)
-				if err != nil {
-					log.Error().
-						Err(err).
-						Str("node", node.Hostname).
-						Str("name", node.GivenName).
-						Msg("🤮 Cannot expire node")
-				} else {
-					log.Info().
-						Str("node", node.Hostname).
-						Str("name", node.GivenName).
-						Msg("Node successfully expired")
-				}
 			}
 		}
+	}
 
-		if len(expired) > 0 {
+	expired := make([]*tailcfg.PeerChange, len(expiredNodes))
-			hsdb.notifier.NotifyAll(types.StateUpdate{
+	for idx, node := range expiredNodes {
-				Type:    types.StatePeerRemoved,
+		expired[idx] = &tailcfg.PeerChange{
-				Removed: expired,
+			NodeID:    tailcfg.NodeID(node.ID),
-			})
+			KeyExpiry: &started,
+		}
+	}
+
+	// Inform the peers of a node with a lightweight update.
+	stateUpdate := types.StateUpdate{
+		Type:          types.StatePeerChangedPatch,
+		ChangePatches: expired,
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyAll(stateUpdate)
+	}
+
+	// Inform the node itself that it has expired.
+	for _, node := range expiredNodes {
+		stateSelfUpdate := types.StateUpdate{
+			Type:        types.StateSelfUpdate,
+			ChangeNodes: types.Nodes{node},
+		}
+		if stateSelfUpdate.Valid() {
+			hsdb.notifier.NotifyByMachineKey(stateSelfUpdate, node.MachineKey)
 		}
 	}
 

hscontrol/db/node_test.go

@@ -12,6 +12,7 @@ import (
 	"github.com/juanfont/headscale/hscontrol/types"
 	"github.com/juanfont/headscale/hscontrol/util"
 	"gopkg.in/check.v1"
+	"tailscale.com/tailcfg"
 	"tailscale.com/types/key"
 )
 
@@ -25,11 +26,13 @@ func (s *Suite) TestGetNode(c *check.C) {
 	_, err = db.GetNode("test", "testnode")
 	c.Assert(err, check.NotNil)
 
+	nodeKey := key.NewNode()
+	machineKey := key.NewMachine()
+
 	node := &types.Node{
 		ID:             0,
-		MachineKey:     "foo",
+		MachineKey:     machineKey.Public(),
-		NodeKey:        "bar",
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -51,11 +54,13 @@ func (s *Suite) TestGetNodeByID(c *check.C) {
 	_, err = db.GetNodeByID(0)
 	c.Assert(err, check.NotNil)
 
+	nodeKey := key.NewNode()
+	machineKey := key.NewMachine()
+
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
+		MachineKey:     machineKey.Public(),
-		NodeKey:        "bar",
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -82,9 +87,8 @@ func (s *Suite) TestGetNodeByNodeKey(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     util.MachinePublicKeyStripPrefix(machineKey.Public()),
+		MachineKey:     machineKey.Public(),
-		NodeKey:        util.NodePublicKeyStripPrefix(nodeKey.Public()),
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -113,9 +117,8 @@ func (s *Suite) TestGetNodeByAnyNodeKey(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     util.MachinePublicKeyStripPrefix(machineKey.Public()),
+		MachineKey:     machineKey.Public(),
-		NodeKey:        util.NodePublicKeyStripPrefix(nodeKey.Public()),
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -130,11 +133,14 @@ func (s *Suite) TestGetNodeByAnyNodeKey(c *check.C) {
 func (s *Suite) TestHardDeleteNode(c *check.C) {
 	user, err := db.CreateUser("test")
 	c.Assert(err, check.IsNil)
+
+	nodeKey := key.NewNode()
+	machineKey := key.NewMachine()
+
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
+		MachineKey:     machineKey.Public(),
-		NodeKey:        "bar",
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "testnode3",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -160,11 +166,13 @@ func (s *Suite) TestListPeers(c *check.C) {
 	c.Assert(err, check.NotNil)
 
 	for index := 0; index <= 10; index++ {
+		nodeKey := key.NewNode()
+		machineKey := key.NewMachine()
+
 		node := types.Node{
 			ID:             uint64(index),
-			MachineKey:     "foo" + strconv.Itoa(index),
+			MachineKey:     machineKey.Public(),
-			NodeKey:        "bar" + strconv.Itoa(index),
+			NodeKey:        nodeKey.Public(),
-			DiscoKey:       "faa" + strconv.Itoa(index),
 			Hostname:       "testnode" + strconv.Itoa(index),
 			UserID:         user.ID,
 			RegisterMethod: util.RegisterMethodAuthKey,
@@ -205,11 +213,13 @@ func (s *Suite) TestGetACLFilteredPeers(c *check.C) {
 	c.Assert(err, check.NotNil)
 
 	for index := 0; index <= 10; index++ {
+		nodeKey := key.NewNode()
+		machineKey := key.NewMachine()
+
 		node := types.Node{
 			ID:         uint64(index),
-			MachineKey: "foo" + strconv.Itoa(index),
+			MachineKey: machineKey.Public(),
-			NodeKey:    "bar" + strconv.Itoa(index),
+			NodeKey:    nodeKey.Public(),
-			DiscoKey:   "faa" + strconv.Itoa(index),
 			IPAddresses: types.NodeAddresses{
 				netip.MustParseAddr(fmt.Sprintf("100.64.0.%v", strconv.Itoa(index+1))),
 			},
@@ -288,11 +298,13 @@ func (s *Suite) TestExpireNode(c *check.C) {
 	_, err = db.GetNode("test", "testnode")
 	c.Assert(err, check.NotNil)
 
+	nodeKey := key.NewNode()
+	machineKey := key.NewMachine()
+
 	node := &types.Node{
 		ID:             0,
-		MachineKey:     "foo",
+		MachineKey:     machineKey.Public(),
-		NodeKey:        "bar",
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -345,11 +357,15 @@ func (s *Suite) TestGenerateGivenName(c *check.C) {
 	_, err = db.GetNode("user-1", "testnode")
 	c.Assert(err, check.NotNil)
 
+	nodeKey := key.NewNode()
+	machineKey := key.NewMachine()
+
+	machineKey2 := key.NewMachine()
+
 	node := &types.Node{
 		ID:             0,
-		MachineKey:     "node-key-1",
+		MachineKey:     machineKey.Public(),
-		NodeKey:        "node-key-1",
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "disco-key-1",
 		Hostname:       "hostname-1",
 		GivenName:      "hostname-1",
 		UserID:         user1.ID,
@@ -358,25 +374,20 @@ func (s *Suite) TestGenerateGivenName(c *check.C) {
 	}
 	db.db.Save(node)
 
-	givenName, err := db.GenerateGivenName("node-key-2", "hostname-2")
+	givenName, err := db.GenerateGivenName(machineKey2.Public(), "hostname-2")
 	comment := check.Commentf("Same user, unique nodes, unique hostnames, no conflict")
 	c.Assert(err, check.IsNil, comment)
 	c.Assert(givenName, check.Equals, "hostname-2", comment)
 
-	givenName, err = db.GenerateGivenName("node-key-1", "hostname-1")
+	givenName, err = db.GenerateGivenName(machineKey.Public(), "hostname-1")
 	comment = check.Commentf("Same user, same node, same hostname, no conflict")
 	c.Assert(err, check.IsNil, comment)
 	c.Assert(givenName, check.Equals, "hostname-1", comment)
 
-	givenName, err = db.GenerateGivenName("node-key-2", "hostname-1")
+	givenName, err = db.GenerateGivenName(machineKey2.Public(), "hostname-1")
 	comment = check.Commentf("Same user, unique nodes, same hostname, conflict")
 	c.Assert(err, check.IsNil, comment)
 	c.Assert(givenName, check.Matches, fmt.Sprintf("^hostname-1-[a-z0-9]{%d}$", NodeGivenNameHashLength), comment)
-
-	givenName, err = db.GenerateGivenName("node-key-2", "hostname-1")
-	comment = check.Commentf("Unique users, unique nodes, same hostname, conflict")
-	c.Assert(err, check.IsNil, comment)
-	c.Assert(givenName, check.Matches, fmt.Sprintf("^hostname-1-[a-z0-9]{%d}$", NodeGivenNameHashLength), comment)
 }
 
 func (s *Suite) TestSetTags(c *check.C) {
@@ -389,11 +400,13 @@ func (s *Suite) TestSetTags(c *check.C) {
 	_, err = db.GetNode("test", "testnode")
 	c.Assert(err, check.NotNil)
 
+	nodeKey := key.NewNode()
+	machineKey := key.NewMachine()
+
 	node := &types.Node{
 		ID:             0,
-		MachineKey:     "foo",
+		MachineKey:     machineKey.Public(),
-		NodeKey:        "bar",
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -565,6 +578,7 @@ func (s *Suite) TestAutoApproveRoutes(c *check.C) {
 	c.Assert(err, check.IsNil)
 
 	nodeKey := key.NewNode()
+	machineKey := key.NewMachine()
 
 	defaultRouteV4 := netip.MustParsePrefix("0.0.0.0/0")
 	defaultRouteV6 := netip.MustParsePrefix("::/0")
@@ -574,14 +588,13 @@ func (s *Suite) TestAutoApproveRoutes(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
+		MachineKey:     machineKey.Public(),
-		NodeKey:        util.NodePublicKeyStripPrefix(nodeKey.Public()),
+		NodeKey:        nodeKey.Public(),
-		DiscoKey:       "faa",
 		Hostname:       "test",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
 		AuthKeyID:      uint(pak.ID),
-		HostInfo: types.HostInfo{
+		Hostinfo: &tailcfg.Hostinfo{
 			RequestTags: []string{"tag:exit"},
 			RoutableIPs: []netip.Prefix{defaultRouteV4, defaultRouteV6, route1, route2},
 		},
@@ -590,8 +603,9 @@ func (s *Suite) TestAutoApproveRoutes(c *check.C) {
 
 	db.db.Save(&node)
 
-	err = db.SaveNodeRoutes(&node)
+	sendUpdate, err := db.SaveNodeRoutes(&node)
 	c.Assert(err, check.IsNil)
+	c.Assert(sendUpdate, check.Equals, false)
 
 	node0ByID, err := db.GetNodeByID(0)
 	c.Assert(err, check.IsNil)

hscontrol/db/preauth_keys_test.go

@@ -77,9 +77,6 @@ func (*Suite) TestAlreadyUsedKey(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "testest",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -101,9 +98,6 @@ func (*Suite) TestReusableBeingUsedKey(c *check.C) {
 
 	node := types.Node{
 		ID:             1,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "testest",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -138,9 +132,6 @@ func (*Suite) TestEphemeralKey(c *check.C) {
 	now := time.Now().Add(-time.Second * 30)
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "testest",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -207,5 +198,5 @@ func (*Suite) TestPreAuthKeyACLTags(c *check.C) {
 
 	listedPaks, err := db.ListPreAuthKeys("test8")
 	c.Assert(err, check.IsNil)
-	c.Assert(listedPaks[0].Proto().AclTags, check.DeepEquals, tags)
+	c.Assert(listedPaks[0].Proto().GetAclTags(), check.DeepEquals, tags)
 }

hscontrol/db/routes.go

@@ -7,7 +7,9 @@ import (
 	"github.com/juanfont/headscale/hscontrol/policy"
 	"github.com/juanfont/headscale/hscontrol/types"
 	"github.com/rs/zerolog/log"
+	"github.com/samber/lo"
 	"gorm.io/gorm"
+	"tailscale.com/types/key"
 )
 
 var ErrRouteIsNotAvailable = errors.New("route is not available")
@@ -21,7 +23,38 @@ func (hsdb *HSDatabase) GetRoutes() (types.Routes, error) {
 
 func (hsdb *HSDatabase) getRoutes() (types.Routes, error) {
 	var routes types.Routes
-	err := hsdb.db.Preload("Node").Find(&routes).Error
+	err := hsdb.db.
+		Preload("Node").
+		Preload("Node.User").
+		Find(&routes).Error
+	if err != nil {
+		return nil, err
+	}
+
+	return routes, nil
+}
+
+func (hsdb *HSDatabase) getAdvertisedAndEnabledRoutes() (types.Routes, error) {
+	var routes types.Routes
+	err := hsdb.db.
+		Preload("Node").
+		Preload("Node.User").
+		Where("advertised = ? AND enabled = ?", true, true).
+		Find(&routes).Error
+	if err != nil {
+		return nil, err
+	}
+
+	return routes, nil
+}
+
+func (hsdb *HSDatabase) getRoutesByPrefix(pref netip.Prefix) (types.Routes, error) {
+	var routes types.Routes
+	err := hsdb.db.
+		Preload("Node").
+		Preload("Node.User").
+		Where("prefix = ?", types.IPPrefix(pref)).
+		Find(&routes).Error
 	if err != nil {
 		return nil, err
 	}
@@ -40,6 +73,7 @@ func (hsdb *HSDatabase) getNodeAdvertisedRoutes(node *types.Node) (types.Routes,
 	var routes types.Routes
 	err := hsdb.db.
 		Preload("Node").
+		Preload("Node.User").
 		Where("node_id = ? AND advertised = true", node.ID).
 		Find(&routes).Error
 	if err != nil {
@@ -60,6 +94,7 @@ func (hsdb *HSDatabase) getNodeRoutes(node *types.Node) (types.Routes, error) {
 	var routes types.Routes
 	err := hsdb.db.
 		Preload("Node").
+		Preload("Node.User").
 		Where("node_id = ?", node.ID).
 		Find(&routes).Error
 	if err != nil && !errors.Is(err, gorm.ErrRecordNotFound) {
@@ -78,7 +113,10 @@ func (hsdb *HSDatabase) GetRoute(id uint64) (*types.Route, error) {
 
 func (hsdb *HSDatabase) getRoute(id uint64) (*types.Route, error) {
 	var route types.Route
-	err := hsdb.db.Preload("Node").First(&route, id).Error
+	err := hsdb.db.
+		Preload("Node").
+		Preload("Node.User").
+		First(&route, id).Error
 	if err != nil {
 		return nil, err
 	}
@@ -122,37 +160,61 @@ func (hsdb *HSDatabase) DisableRoute(id uint64) error {
 		return err
 	}
 
+	var routes types.Routes
+	node := route.Node
+
 	// Tailscale requires both IPv4 and IPv6 exit routes to
 	// be enabled at the same time, as per
 	// https://github.com/juanfont/headscale/issues/804#issuecomment-1399314002
 	if !route.IsExitRoute() {
+		err = hsdb.failoverRouteWithNotify(route)
+		if err != nil {
+			return err
+		}
+
 		route.Enabled = false
 		route.IsPrimary = false
 		err = hsdb.db.Save(route).Error
 		if err != nil {
 			return err
 		}
+	} else {
+		routes, err = hsdb.getNodeRoutes(&node)
+		if err != nil {
+			return err
+		}
 
-		return hsdb.handlePrimarySubnetFailover()
+		for i := range routes {
-	}
+			if routes[i].IsExitRoute() {
-
+				routes[i].Enabled = false
-	routes, err := hsdb.getNodeRoutes(&route.Node)
+				routes[i].IsPrimary = false
-	if err != nil {
+				err = hsdb.db.Save(&routes[i]).Error
-		return err
+				if err != nil {
-	}
+					return err
-
+				}
-	for i := range routes {
-		if routes[i].IsExitRoute() {
-			routes[i].Enabled = false
-			routes[i].IsPrimary = false
-			err = hsdb.db.Save(&routes[i]).Error
-			if err != nil {
-				return err
 			}
 		}
 	}
 
-	return hsdb.handlePrimarySubnetFailover()
+	if routes == nil {
+		routes, err = hsdb.getNodeRoutes(&node)
+		if err != nil {
+			return err
+		}
+	}
+
+	node.Routes = routes
+
+	stateUpdate := types.StateUpdate{
+		Type:        types.StatePeerChanged,
+		ChangeNodes: types.Nodes{&node},
+		Message:     "called from db.DisableRoute",
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyAll(stateUpdate)
+	}
+
+	return nil
 }
 
 func (hsdb *HSDatabase) DeleteRoute(id uint64) error {
@@ -164,34 +226,58 @@ func (hsdb *HSDatabase) DeleteRoute(id uint64) error {
 		return err
 	}
 
+	var routes types.Routes
+	node := route.Node
+
 	// Tailscale requires both IPv4 and IPv6 exit routes to
 	// be enabled at the same time, as per
 	// https://github.com/juanfont/headscale/issues/804#issuecomment-1399314002
 	if !route.IsExitRoute() {
+		err := hsdb.failoverRouteWithNotify(route)
+		if err != nil {
+			return nil
+		}
+
 		if err := hsdb.db.Unscoped().Delete(&route).Error; err != nil {
 			return err
 		}
+	} else {
+		routes, err := hsdb.getNodeRoutes(&node)
+		if err != nil {
+			return err
+		}
 
-		return hsdb.handlePrimarySubnetFailover()
+		routesToDelete := types.Routes{}
-	}
+		for _, r := range routes {
+			if r.IsExitRoute() {
+				routesToDelete = append(routesToDelete, r)
+			}
+		}
 
-	routes, err := hsdb.getNodeRoutes(&route.Node)
+		if err := hsdb.db.Unscoped().Delete(&routesToDelete).Error; err != nil {
-	if err != nil {
+			return err
-		return err
-	}
-
-	routesToDelete := types.Routes{}
-	for _, r := range routes {
-		if r.IsExitRoute() {
-			routesToDelete = append(routesToDelete, r)
 		}
 	}
 
-	if err := hsdb.db.Unscoped().Delete(&routesToDelete).Error; err != nil {
+	if routes == nil {
-		return err
+		routes, err = hsdb.getNodeRoutes(&node)
+		if err != nil {
+			return err
+		}
 	}
 
-	return hsdb.handlePrimarySubnetFailover()
+	node.Routes = routes
+
+	stateUpdate := types.StateUpdate{
+		Type:        types.StatePeerChanged,
+		ChangeNodes: types.Nodes{&node},
+		Message:     "called from db.DeleteRoute",
+	}
+	if stateUpdate.Valid() {
+		hsdb.notifier.NotifyAll(stateUpdate)
+	}
+
+	return nil
 }
 
 func (hsdb *HSDatabase) deleteNodeRoutes(node *types.Node) error {
@@ -204,9 +290,13 @@ func (hsdb *HSDatabase) deleteNodeRoutes(node *types.Node) error {
 		if err := hsdb.db.Unscoped().Delete(&routes[i]).Error; err != nil {
 			return err
 		}
+
+		// TODO(kradalby): This is a bit too aggressive, we could probably
+		// figure out which routes needs to be failed over rather than all.
+		hsdb.failoverRouteWithNotify(&routes[i])
 	}
 
-	return hsdb.handlePrimarySubnetFailover()
+	return nil
 }
 
 // isUniquePrefix returns if there is another node providing the same route already.
@@ -259,22 +349,26 @@ func (hsdb *HSDatabase) GetNodePrimaryRoutes(node *types.Node) (types.Routes, er
 
 // SaveNodeRoutes takes a node and updates the database with
 // the new routes.
-func (hsdb *HSDatabase) SaveNodeRoutes(node *types.Node) error {
+// It returns a bool wheter an update should be sent as the
+// saved route impacts nodes.
+func (hsdb *HSDatabase) SaveNodeRoutes(node *types.Node) (bool, error) {
 	hsdb.mu.Lock()
 	defer hsdb.mu.Unlock()
 
 	return hsdb.saveNodeRoutes(node)
 }
 
-func (hsdb *HSDatabase) saveNodeRoutes(node *types.Node) error {
+func (hsdb *HSDatabase) saveNodeRoutes(node *types.Node) (bool, error) {
+	sendUpdate := false
+
 	currentRoutes := types.Routes{}
 	err := hsdb.db.Where("node_id = ?", node.ID).Find(&currentRoutes).Error
 	if err != nil {
-		return err
+		return sendUpdate, err
 	}
 
 	advertisedRoutes := map[netip.Prefix]bool{}
-	for _, prefix := range node.HostInfo.RoutableIPs {
+	for _, prefix := range node.Hostinfo.RoutableIPs {
 		advertisedRoutes[prefix] = false
 	}
 
@@ -290,7 +384,14 @@ func (hsdb *HSDatabase) saveNodeRoutes(node *types.Node) error {
 				currentRoutes[pos].Advertised = true
 				err := hsdb.db.Save(&currentRoutes[pos]).Error
 				if err != nil {
-					return err
+					return sendUpdate, err
+				}
+
+				// If a route that is newly "saved" is already
+				// enabled, set sendUpdate to true as it is now
+				// available.
+				if route.Enabled {
+					sendUpdate = true
 				}
 			}
 			advertisedRoutes[netip.Prefix(route.Prefix)] = true
@@ -299,7 +400,7 @@ func (hsdb *HSDatabase) saveNodeRoutes(node *types.Node) error {
 			currentRoutes[pos].Enabled = false
 			err := hsdb.db.Save(&currentRoutes[pos]).Error
 			if err != nil {
-				return err
+				return sendUpdate, err
 			}
 		}
 	}
@@ -314,7 +415,41 @@ func (hsdb *HSDatabase) saveNodeRoutes(node *types.Node) error {
 			}
 			err := hsdb.db.Create(&route).Error
 			if err != nil {
-				return err
+				return sendUpdate, err
+			}
+		}
+	}
+
+	return sendUpdate, nil
+}
+
+// EnsureFailoverRouteIsAvailable takes a node and checks if the node's route
+// currently have a functioning host that exposes the network.
+func (hsdb *HSDatabase) EnsureFailoverRouteIsAvailable(node *types.Node) error {
+	nodeRoutes, err := hsdb.getNodeRoutes(node)
+	if err != nil {
+		return nil
+	}
+
+	for _, nodeRoute := range nodeRoutes {
+		routes, err := hsdb.getRoutesByPrefix(netip.Prefix(nodeRoute.Prefix))
+		if err != nil {
+			return err
+		}
+
+		for _, route := range routes {
+			if route.IsPrimary {
+				// if we have a primary route, and the node is connected
+				// nothing needs to be done.
+				if hsdb.notifier.IsConnected(route.Node.MachineKey) {
+					continue
+				}
+
+				// if not, we need to failover the route
+				err := hsdb.failoverRouteWithNotify(&route)
+				if err != nil {
+					return err
+				}
 			}
 		}
 	}
@@ -322,133 +457,181 @@ func (hsdb *HSDatabase) saveNodeRoutes(node *types.Node) error {
 	return nil
 }
 
-func (hsdb *HSDatabase) HandlePrimarySubnetFailover() error {
+func (hsdb *HSDatabase) FailoverNodeRoutesWithNotify(node *types.Node) error {
-	hsdb.mu.Lock()
+	routes, err := hsdb.getNodeRoutes(node)
-	defer hsdb.mu.Unlock()
+	if err != nil {
-
+		return nil
-	return hsdb.handlePrimarySubnetFailover()
-}
-
-func (hsdb *HSDatabase) handlePrimarySubnetFailover() error {
-	// first, get all the enabled routes
-	var routes types.Routes
-	err := hsdb.db.
-		Preload("Node").
-		Where("advertised = ? AND enabled = ?", true, true).
-		Find(&routes).Error
-	if err != nil && !errors.Is(err, gorm.ErrRecordNotFound) {
-		log.Error().Err(err).Msg("error getting routes")
 	}
 
-	changedNodes := make(types.Nodes, 0)
+	var changedKeys []key.MachinePublic
-	for pos, route := range routes {
+
-		if route.IsExitRoute() {
+	for _, route := range routes {
+		changed, err := hsdb.failoverRoute(&route)
+		if err != nil {
+			return err
+		}
+
+		changedKeys = append(changedKeys, changed...)
+	}
+
+	changedKeys = lo.Uniq(changedKeys)
+
+	var nodes types.Nodes
+
+	for _, key := range changedKeys {
+		node, err := hsdb.GetNodeByMachineKey(key)
+		if err != nil {
+			return err
+		}
+
+		nodes = append(nodes, node)
+	}
+
+	if nodes != nil {
+		stateUpdate := types.StateUpdate{
+			Type:        types.StatePeerChanged,
+			ChangeNodes: nodes,
+			Message:     "called from db.FailoverNodeRoutesWithNotify",
+		}
+		if stateUpdate.Valid() {
+			hsdb.notifier.NotifyAll(stateUpdate)
+		}
+	}
+
+	return nil
+}
+
+func (hsdb *HSDatabase) failoverRouteWithNotify(r *types.Route) error {
+	changedKeys, err := hsdb.failoverRoute(r)
+	if err != nil {
+		return err
+	}
+
+	if len(changedKeys) == 0 {
+		return nil
+	}
+
+	var nodes types.Nodes
+
+	log.Trace().
+		Str("hostname", r.Node.Hostname).
+		Msg("loading machines with new primary routes from db")
+
+	for _, key := range changedKeys {
+		node, err := hsdb.getNodeByMachineKey(key)
+		if err != nil {
+			return err
+		}
+
+		nodes = append(nodes, node)
+	}
+
+	log.Trace().
+		Str("hostname", r.Node.Hostname).
+		Msg("notifying peers about primary route change")
+
+	if nodes != nil {
+		stateUpdate := types.StateUpdate{
+			Type:        types.StatePeerChanged,
+			ChangeNodes: nodes,
+			Message:     "called from db.failoverRouteWithNotify",
+		}
+		if stateUpdate.Valid() {
+			hsdb.notifier.NotifyAll(stateUpdate)
+		}
+	}
+
+	log.Trace().
+		Str("hostname", r.Node.Hostname).
+		Msg("notified peers about primary route change")
+
+	return nil
+}
+
+// failoverRoute takes a route that is no longer available,
+// this can be either from:
+// - being disabled
+// - being deleted
+// - host going offline
+//
+// and tries to find a new route to take over its place.
+// If the given route was not primary, it returns early.
+func (hsdb *HSDatabase) failoverRoute(r *types.Route) ([]key.MachinePublic, error) {
+	if r == nil {
+		return nil, nil
+	}
+
+	// This route is not a primary route, and it isnt
+	// being served to nodes.
+	if !r.IsPrimary {
+		return nil, nil
+	}
+
+	// We do not have to failover exit nodes
+	if r.IsExitRoute() {
+		return nil, nil
+	}
+
+	routes, err := hsdb.getRoutesByPrefix(netip.Prefix(r.Prefix))
+	if err != nil {
+		return nil, err
+	}
+
+	var newPrimary *types.Route
+
+	// Find a new suitable route
+	for idx, route := range routes {
+		if r.ID == route.ID {
 			continue
 		}
 
-		node := &route.Node
+		if hsdb.notifier.IsConnected(route.Node.MachineKey) {
-
+			newPrimary = &routes[idx]
-		if !route.IsPrimary {
+			break
-			_, err := hsdb.getPrimaryRoute(netip.Prefix(route.Prefix))
-			if hsdb.isUniquePrefix(route) || errors.Is(err, gorm.ErrRecordNotFound) {
-				log.Info().
-					Str("prefix", netip.Prefix(route.Prefix).String()).
-					Str("node", route.Node.GivenName).
-					Msg("Setting primary route")
-				routes[pos].IsPrimary = true
-				err := hsdb.db.Save(&routes[pos]).Error
-				if err != nil {
-					log.Error().Err(err).Msg("error marking route as primary")
-
-					return err
-				}
-
-				changedNodes = append(changedNodes, node)
-
-				continue
-			}
-		}
-
-		if route.IsPrimary {
-			if route.Node.IsOnline() {
-				continue
-			}
-
-			// node offline, find a new primary
-			log.Info().
-				Str("node", route.Node.Hostname).
-				Str("prefix", netip.Prefix(route.Prefix).String()).
-				Msgf("node offline, finding a new primary subnet")
-
-			// find a new primary route
-			var newPrimaryRoutes types.Routes
-			err := hsdb.db.
-				Preload("Node").
-				Where("prefix = ? AND node_id != ? AND advertised = ? AND enabled = ?",
-					route.Prefix,
-					route.NodeID,
-					true, true).
-				Find(&newPrimaryRoutes).Error
-			if err != nil && !errors.Is(err, gorm.ErrRecordNotFound) {
-				log.Error().Err(err).Msg("error finding new primary route")
-
-				return err
-			}
-
-			var newPrimaryRoute *types.Route
-			for pos, r := range newPrimaryRoutes {
-				if r.Node.IsOnline() {
-					newPrimaryRoute = &newPrimaryRoutes[pos]
-
-					break
-				}
-			}
-
-			if newPrimaryRoute == nil {
-				log.Warn().
-					Str("node", route.Node.Hostname).
-					Str("prefix", netip.Prefix(route.Prefix).String()).
-					Msgf("no alternative primary route found")
-
-				continue
-			}
-
-			log.Info().
-				Str("old_node", route.Node.Hostname).
-				Str("prefix", netip.Prefix(route.Prefix).String()).
-				Str("new_node", newPrimaryRoute.Node.Hostname).
-				Msgf("found new primary route")
-
-			// disable the old primary route
-			routes[pos].IsPrimary = false
-			err = hsdb.db.Save(&routes[pos]).Error
-			if err != nil {
-				log.Error().Err(err).Msg("error disabling old primary route")
-
-				return err
-			}
-
-			// enable the new primary route
-			newPrimaryRoute.IsPrimary = true
-			err = hsdb.db.Save(&newPrimaryRoute).Error
-			if err != nil {
-				log.Error().Err(err).Msg("error enabling new primary route")
-
-				return err
-			}
-
-			changedNodes = append(changedNodes, node)
 		}
 	}
 
-	if len(changedNodes) > 0 {
+	// If a new route was not found/available,
-		hsdb.notifier.NotifyAll(types.StateUpdate{
+	// return with an error.
-			Type:    types.StatePeerChanged,
+	// We do not want to update the database as
-			Changed: changedNodes,
+	// the one currently marked as primary is the
-		})
+	// best we got.
+	if newPrimary == nil {
+		return nil, nil
 	}
 
-	return nil
+	log.Trace().
+		Str("hostname", newPrimary.Node.Hostname).
+		Msg("found new primary, updating db")
+
+	// Remove primary from the old route
+	r.IsPrimary = false
+	err = hsdb.db.Save(&r).Error
+	if err != nil {
+		log.Error().Err(err).Msg("error disabling new primary route")
+
+		return nil, err
+	}
+
+	log.Trace().
+		Str("hostname", newPrimary.Node.Hostname).
+		Msg("removed primary from old route")
+
+	// Set primary for the new primary
+	newPrimary.IsPrimary = true
+	err = hsdb.db.Save(&newPrimary).Error
+	if err != nil {
+		log.Error().Err(err).Msg("error enabling new primary route")
+
+		return nil, err
+	}
+
+	log.Trace().
+		Str("hostname", newPrimary.Node.Hostname).
+		Msg("set primary to new route")
+
+	// Return a list of the machinekeys of the changed nodes.
+	return []key.MachinePublic{r.Node.MachineKey, newPrimary.Node.MachineKey}, nil
 }
 
 // EnableAutoApprovedRoutes enables any routes advertised by a node that match the ACL autoApprovers policy.

hscontrol/db/routes_test.go

@@ -2,12 +2,19 @@ package db
 
 import (
 	"net/netip"
+	"os"
+	"testing"
 	"time"
 
+	"github.com/google/go-cmp/cmp"
+	"github.com/juanfont/headscale/hscontrol/notifier"
 	"github.com/juanfont/headscale/hscontrol/types"
 	"github.com/juanfont/headscale/hscontrol/util"
+	"github.com/stretchr/testify/assert"
 	"gopkg.in/check.v1"
+	"gorm.io/gorm"
 	"tailscale.com/tailcfg"
+	"tailscale.com/types/key"
 )
 
 func (s *Suite) TestGetRoutes(c *check.C) {
@@ -29,19 +36,17 @@ func (s *Suite) TestGetRoutes(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "test_get_route_node",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
 		AuthKeyID:      uint(pak.ID),
-		HostInfo:       types.HostInfo(hostInfo),
+		Hostinfo:       &hostInfo,
 	}
 	db.db.Save(&node)
 
-	err = db.SaveNodeRoutes(&node)
+	su, err := db.SaveNodeRoutes(&node)
 	c.Assert(err, check.IsNil)
+	c.Assert(su, check.Equals, false)
 
 	advertisedRoutes, err := db.GetAdvertisedRoutes(&node)
 	c.Assert(err, check.IsNil)
@@ -80,19 +85,17 @@ func (s *Suite) TestGetEnableRoutes(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "test_enable_route_node",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
 		AuthKeyID:      uint(pak.ID),
-		HostInfo:       types.HostInfo(hostInfo),
+		Hostinfo:       &hostInfo,
 	}
 	db.db.Save(&node)
 
-	err = db.SaveNodeRoutes(&node)
+	sendUpdate, err := db.SaveNodeRoutes(&node)
 	c.Assert(err, check.IsNil)
+	c.Assert(sendUpdate, check.Equals, false)
 
 	availableRoutes, err := db.GetAdvertisedRoutes(&node)
 	c.Assert(err, check.IsNil)
@@ -154,19 +157,17 @@ func (s *Suite) TestIsUniquePrefix(c *check.C) {
 	}
 	node1 := types.Node{
 		ID:             1,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "test_enable_route_node",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
 		AuthKeyID:      uint(pak.ID),
-		HostInfo:       types.HostInfo(hostInfo1),
+		Hostinfo:       &hostInfo1,
 	}
 	db.db.Save(&node1)
 
-	err = db.SaveNodeRoutes(&node1)
+	sendUpdate, err := db.SaveNodeRoutes(&node1)
 	c.Assert(err, check.IsNil)
+	c.Assert(sendUpdate, check.Equals, false)
 
 	err = db.enableRoutes(&node1, route.String())
 	c.Assert(err, check.IsNil)
@@ -179,19 +180,17 @@ func (s *Suite) TestIsUniquePrefix(c *check.C) {
 	}
 	node2 := types.Node{
 		ID:             2,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "test_enable_route_node",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
 		AuthKeyID:      uint(pak.ID),
-		HostInfo:       types.HostInfo(hostInfo2),
+		Hostinfo:       &hostInfo2,
 	}
 	db.db.Save(&node2)
 
-	err = db.SaveNodeRoutes(&node2)
+	sendUpdate, err = db.SaveNodeRoutes(&node2)
 	c.Assert(err, check.IsNil)
+	c.Assert(sendUpdate, check.Equals, false)
 
 	err = db.enableRoutes(&node2, route2.String())
 	c.Assert(err, check.IsNil)
@@ -213,148 +212,6 @@ func (s *Suite) TestIsUniquePrefix(c *check.C) {
 	c.Assert(len(routes), check.Equals, 0)
 }
 
-func (s *Suite) TestSubnetFailover(c *check.C) {
-	user, err := db.CreateUser("test")
-	c.Assert(err, check.IsNil)
-
-	pak, err := db.CreatePreAuthKey(user.Name, false, false, nil, nil)
-	c.Assert(err, check.IsNil)
-
-	_, err = db.GetNode("test", "test_enable_route_node")
-	c.Assert(err, check.NotNil)
-
-	prefix, err := netip.ParsePrefix(
-		"10.0.0.0/24",
-	)
-	c.Assert(err, check.IsNil)
-
-	prefix2, err := netip.ParsePrefix(
-		"150.0.10.0/25",
-	)
-	c.Assert(err, check.IsNil)
-
-	hostInfo1 := tailcfg.Hostinfo{
-		RoutableIPs: []netip.Prefix{prefix, prefix2},
-	}
-
-	now := time.Now()
-	node1 := types.Node{
-		ID:             1,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
-		Hostname:       "test_enable_route_node",
-		UserID:         user.ID,
-		RegisterMethod: util.RegisterMethodAuthKey,
-		AuthKeyID:      uint(pak.ID),
-		HostInfo:       types.HostInfo(hostInfo1),
-		LastSeen:       &now,
-	}
-	db.db.Save(&node1)
-
-	err = db.SaveNodeRoutes(&node1)
-	c.Assert(err, check.IsNil)
-
-	err = db.enableRoutes(&node1, prefix.String())
-	c.Assert(err, check.IsNil)
-
-	err = db.enableRoutes(&node1, prefix2.String())
-	c.Assert(err, check.IsNil)
-
-	err = db.HandlePrimarySubnetFailover()
-	c.Assert(err, check.IsNil)
-
-	enabledRoutes1, err := db.GetEnabledRoutes(&node1)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(enabledRoutes1), check.Equals, 2)
-
-	route, err := db.getPrimaryRoute(prefix)
-	c.Assert(err, check.IsNil)
-	c.Assert(route.NodeID, check.Equals, node1.ID)
-
-	hostInfo2 := tailcfg.Hostinfo{
-		RoutableIPs: []netip.Prefix{prefix2},
-	}
-	node2 := types.Node{
-		ID:             2,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
-		Hostname:       "test_enable_route_node",
-		UserID:         user.ID,
-		RegisterMethod: util.RegisterMethodAuthKey,
-		AuthKeyID:      uint(pak.ID),
-		HostInfo:       types.HostInfo(hostInfo2),
-		LastSeen:       &now,
-	}
-	db.db.Save(&node2)
-
-	err = db.saveNodeRoutes(&node2)
-	c.Assert(err, check.IsNil)
-
-	err = db.enableRoutes(&node2, prefix2.String())
-	c.Assert(err, check.IsNil)
-
-	err = db.HandlePrimarySubnetFailover()
-	c.Assert(err, check.IsNil)
-
-	enabledRoutes1, err = db.GetEnabledRoutes(&node1)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(enabledRoutes1), check.Equals, 2)
-
-	enabledRoutes2, err := db.GetEnabledRoutes(&node2)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(enabledRoutes2), check.Equals, 1)
-
-	routes, err := db.GetNodePrimaryRoutes(&node1)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(routes), check.Equals, 2)
-
-	routes, err = db.GetNodePrimaryRoutes(&node2)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(routes), check.Equals, 0)
-
-	// lets make node1 lastseen 10 mins ago
-	before := now.Add(-10 * time.Minute)
-	node1.LastSeen = &before
-	err = db.db.Save(&node1).Error
-	c.Assert(err, check.IsNil)
-
-	err = db.HandlePrimarySubnetFailover()
-	c.Assert(err, check.IsNil)
-
-	routes, err = db.GetNodePrimaryRoutes(&node1)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(routes), check.Equals, 1)
-
-	routes, err = db.GetNodePrimaryRoutes(&node2)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(routes), check.Equals, 1)
-
-	node2.HostInfo = types.HostInfo(tailcfg.Hostinfo{
-		RoutableIPs: []netip.Prefix{prefix, prefix2},
-	})
-	err = db.db.Save(&node2).Error
-	c.Assert(err, check.IsNil)
-
-	err = db.SaveNodeRoutes(&node2)
-	c.Assert(err, check.IsNil)
-
-	err = db.enableRoutes(&node2, prefix.String())
-	c.Assert(err, check.IsNil)
-
-	err = db.HandlePrimarySubnetFailover()
-	c.Assert(err, check.IsNil)
-
-	routes, err = db.GetNodePrimaryRoutes(&node1)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(routes), check.Equals, 0)
-
-	routes, err = db.GetNodePrimaryRoutes(&node2)
-	c.Assert(err, check.IsNil)
-	c.Assert(len(routes), check.Equals, 2)
-}
-
 func (s *Suite) TestDeleteRoutes(c *check.C) {
 	user, err := db.CreateUser("test")
 	c.Assert(err, check.IsNil)
@@ -382,20 +239,18 @@ func (s *Suite) TestDeleteRoutes(c *check.C) {
 	now := time.Now()
 	node1 := types.Node{
 		ID:             1,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "test_enable_route_node",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
 		AuthKeyID:      uint(pak.ID),
-		HostInfo:       types.HostInfo(hostInfo1),
+		Hostinfo:       &hostInfo1,
 		LastSeen:       &now,
 	}
 	db.db.Save(&node1)
 
-	err = db.SaveNodeRoutes(&node1)
+	sendUpdate, err := db.SaveNodeRoutes(&node1)
 	c.Assert(err, check.IsNil)
+	c.Assert(sendUpdate, check.Equals, false)
 
 	err = db.enableRoutes(&node1, prefix.String())
 	c.Assert(err, check.IsNil)
@@ -413,3 +268,362 @@ func (s *Suite) TestDeleteRoutes(c *check.C) {
 	c.Assert(err, check.IsNil)
 	c.Assert(len(enabledRoutes1), check.Equals, 1)
 }
+
+func TestFailoverRoute(t *testing.T) {
+	ipp := func(s string) types.IPPrefix { return types.IPPrefix(netip.MustParsePrefix(s)) }
+
+	// TODO(kradalby): Count/verify updates
+	var sink chan types.StateUpdate
+
+	go func() {
+		for range sink {
+		}
+	}()
+
+	machineKeys := []key.MachinePublic{
+		key.NewMachine().Public(),
+		key.NewMachine().Public(),
+		key.NewMachine().Public(),
+		key.NewMachine().Public(),
+	}
+
+	tests := []struct {
+		name         string
+		failingRoute types.Route
+		routes       types.Routes
+		want         []key.MachinePublic
+		wantErr      bool
+	}{
+		{
+			name:         "no-route",
+			failingRoute: types.Route{},
+			routes:       types.Routes{},
+			want:         nil,
+			wantErr:      false,
+		},
+		{
+			name: "no-prime",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 1,
+				},
+				Prefix: ipp("10.0.0.0/24"),
+				Node: types.Node{
+					MachineKey: machineKeys[0],
+				},
+				IsPrimary: false,
+			},
+			routes:  types.Routes{},
+			want:    nil,
+			wantErr: false,
+		},
+		{
+			name: "exit-node",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 1,
+				},
+				Prefix: ipp("0.0.0.0/0"),
+				Node: types.Node{
+					MachineKey: machineKeys[0],
+				},
+				IsPrimary: true,
+			},
+			routes:  types.Routes{},
+			want:    nil,
+			wantErr: false,
+		},
+		{
+			name: "no-failover-single-route",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 1,
+				},
+				Prefix: ipp("10.0.0.0/24"),
+				Node: types.Node{
+					MachineKey: machineKeys[0],
+				},
+				IsPrimary: true,
+			},
+			routes: types.Routes{
+				types.Route{
+					Model: gorm.Model{
+						ID: 1,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[0],
+					},
+					IsPrimary: true,
+				},
+			},
+			want:    nil,
+			wantErr: false,
+		},
+		{
+			name: "failover-primary",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 1,
+				},
+				Prefix: ipp("10.0.0.0/24"),
+				Node: types.Node{
+					MachineKey: machineKeys[0],
+				},
+				IsPrimary: true,
+			},
+			routes: types.Routes{
+				types.Route{
+					Model: gorm.Model{
+						ID: 1,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[0],
+					},
+					IsPrimary: true,
+				},
+				types.Route{
+					Model: gorm.Model{
+						ID: 2,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[1],
+					},
+					IsPrimary: false,
+				},
+			},
+			want: []key.MachinePublic{
+				machineKeys[0],
+				machineKeys[1],
+			},
+			wantErr: false,
+		},
+		{
+			name: "failover-none-primary",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 1,
+				},
+				Prefix: ipp("10.0.0.0/24"),
+				Node: types.Node{
+					MachineKey: machineKeys[0],
+				},
+				IsPrimary: false,
+			},
+			routes: types.Routes{
+				types.Route{
+					Model: gorm.Model{
+						ID: 1,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[0],
+					},
+					IsPrimary: true,
+				},
+				types.Route{
+					Model: gorm.Model{
+						ID: 2,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[1],
+					},
+					IsPrimary: false,
+				},
+			},
+			want:    nil,
+			wantErr: false,
+		},
+		{
+			name: "failover-primary-multi-route",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 2,
+				},
+				Prefix: ipp("10.0.0.0/24"),
+				Node: types.Node{
+					MachineKey: machineKeys[1],
+				},
+				IsPrimary: true,
+			},
+			routes: types.Routes{
+				types.Route{
+					Model: gorm.Model{
+						ID: 1,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[0],
+					},
+					IsPrimary: false,
+				},
+				types.Route{
+					Model: gorm.Model{
+						ID: 2,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[1],
+					},
+					IsPrimary: true,
+				},
+				types.Route{
+					Model: gorm.Model{
+						ID: 3,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[2],
+					},
+					IsPrimary: false,
+				},
+			},
+			want: []key.MachinePublic{
+				machineKeys[1],
+				machineKeys[0],
+			},
+			wantErr: false,
+		},
+		{
+			name: "failover-primary-no-online",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 1,
+				},
+				Prefix: ipp("10.0.0.0/24"),
+				Node: types.Node{
+					MachineKey: machineKeys[0],
+				},
+				IsPrimary: true,
+			},
+			routes: types.Routes{
+				types.Route{
+					Model: gorm.Model{
+						ID: 1,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[0],
+					},
+					IsPrimary: true,
+				},
+				// Offline
+				types.Route{
+					Model: gorm.Model{
+						ID: 2,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[3],
+					},
+					IsPrimary: false,
+				},
+			},
+			want:    nil,
+			wantErr: false,
+		},
+		{
+			name: "failover-primary-one-not-online",
+			failingRoute: types.Route{
+				Model: gorm.Model{
+					ID: 1,
+				},
+				Prefix: ipp("10.0.0.0/24"),
+				Node: types.Node{
+					MachineKey: machineKeys[0],
+				},
+				IsPrimary: true,
+			},
+			routes: types.Routes{
+				types.Route{
+					Model: gorm.Model{
+						ID: 1,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[0],
+					},
+					IsPrimary: true,
+				},
+				// Offline
+				types.Route{
+					Model: gorm.Model{
+						ID: 2,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[3],
+					},
+					IsPrimary: false,
+				},
+				types.Route{
+					Model: gorm.Model{
+						ID: 3,
+					},
+					Prefix: ipp("10.0.0.0/24"),
+					Node: types.Node{
+						MachineKey: machineKeys[1],
+					},
+					IsPrimary: true,
+				},
+			},
+			want: []key.MachinePublic{
+				machineKeys[0],
+				machineKeys[1],
+			},
+			wantErr: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			tmpDir, err := os.MkdirTemp("", "failover-db-test")
+			assert.NoError(t, err)
+
+			notif := notifier.NewNotifier()
+
+			db, err = NewHeadscaleDatabase(
+				"sqlite3",
+				tmpDir+"/headscale_test.db",
+				false,
+				notif,
+				[]netip.Prefix{
+					netip.MustParsePrefix("10.27.0.0/23"),
+				},
+				"",
+			)
+			assert.NoError(t, err)
+
+			// Pretend that all the nodes are connected to control
+			for idx, key := range machineKeys {
+				// Pretend one node is offline
+				if idx == 3 {
+					continue
+				}
+
+				notif.AddNode(key, sink)
+			}
+
+			for _, route := range tt.routes {
+				if err := db.db.Save(&route).Error; err != nil {
+					t.Fatalf("failed to create route: %s", err)
+				}
+			}
+
+			got, err := db.failoverRoute(&tt.failingRoute)
+
+			if (err != nil) != tt.wantErr {
+				t.Errorf("failoverRoute() error = %v, wantErr %v", err, tt.wantErr)
+
+				return
+			}
+
+			if diff := cmp.Diff(tt.want, got, util.Comparers...); diff != "" {
+				t.Errorf("failoverRoute() unexpected result (-want +got):\n%s", diff)
+			}
+		})
+	}
+}

hscontrol/db/suite_test.go

@@ -1,6 +1,7 @@
 package db
 
 import (
+	"log"
 	"net/netip"
 	"os"
 	"testing"
@@ -27,19 +28,22 @@ func (s *Suite) SetUpTest(c *check.C) {
 }
 
 func (s *Suite) TearDownTest(c *check.C) {
-	os.RemoveAll(tmpDir)
+	// os.RemoveAll(tmpDir)
 }
 
 func (s *Suite) ResetDB(c *check.C) {
-	if len(tmpDir) != 0 {
+	// if len(tmpDir) != 0 {
-		os.RemoveAll(tmpDir)
+	// 	os.RemoveAll(tmpDir)
-	}
+	// }
+
 	var err error
-	tmpDir, err = os.MkdirTemp("", "autoygg-client-test")
+	tmpDir, err = os.MkdirTemp("", "headscale-db-test-*")
 	if err != nil {
 		c.Fatal(err)
 	}
 
+	log.Printf("database path: %s", tmpDir+"/headscale_test.db")
+
 	db, err = NewHeadscaleDatabase(
 		"sqlite3",
 		tmpDir+"/headscale_test.db",

hscontrol/db/users_test.go

@@ -48,9 +48,6 @@ func (s *Suite) TestDestroyUserErrors(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         user.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,
@@ -103,9 +100,6 @@ func (s *Suite) TestSetMachineUser(c *check.C) {
 
 	node := types.Node{
 		ID:             0,
-		MachineKey:     "foo",
-		NodeKey:        "bar",
-		DiscoKey:       "faa",
 		Hostname:       "testnode",
 		UserID:         oldUser.ID,
 		RegisterMethod: util.RegisterMethodAuthKey,

hscontrol/derp/server/derp_server.go

@@ -13,6 +13,7 @@ import (
 	"time"
 
 	"github.com/juanfont/headscale/hscontrol/types"
+	"github.com/juanfont/headscale/hscontrol/util"
 	"github.com/rs/zerolog/log"
 	"tailscale.com/derp"
 	"tailscale.com/net/stun"
@@ -39,7 +40,7 @@ func NewDERPServer(
 	cfg *types.DERPConfig,
 ) (*DERPServer, error) {
 	log.Trace().Caller().Msg("Creating new embedded DERP server")
-	server := derp.NewServer(derpKey, log.Debug().Msgf) // nolint // zerolinter complains
+	server := derp.NewServer(derpKey, util.TSLogfWrapper()) // nolint // zerolinter complains
 
 	return &DERPServer{
 		serverURL:     serverURL,

hscontrol/grpcv1.go

@@ -172,12 +172,18 @@ func (api headscaleV1APIServer) RegisterNode(
 ) (*v1.RegisterNodeResponse, error) {
 	log.Trace().
 		Str("user", request.GetUser()).
-		Str("node_key", request.GetKey()).
+		Str("machine_key", request.GetKey()).
 		Msg("Registering node")
 
+	var mkey key.MachinePublic
+	err := mkey.UnmarshalText([]byte(request.GetKey()))
+	if err != nil {
+		return nil, err
+	}
+
 	node, err := api.h.db.RegisterNodeFromAuthCallback(
 		api.h.registrationCache,
-		request.GetKey(),
+		mkey,
 		request.GetUser(),
 		nil,
 		util.RegisterMethodCLI,
@@ -198,7 +204,13 @@ func (api headscaleV1APIServer) GetNode(
 		return nil, err
 	}
 
-	return &v1.GetNodeResponse{Node: node.Proto()}, nil
+	resp := node.Proto()
+
+	// Populate the online field based on
+	// currently connected nodes.
+	resp.Online = api.h.nodeNotifier.IsConnected(node.MachineKey)
+
+	return &v1.GetNodeResponse{Node: resp}, nil
 }
 
 func (api headscaleV1APIServer) SetTags(
@@ -327,7 +339,13 @@ func (api headscaleV1APIServer) ListNodes(
 
 		response := make([]*v1.Node, len(nodes))
 		for index, node := range nodes {
-			response[index] = node.Proto()
+			resp := node.Proto()
+
+			// Populate the online field based on
+			// currently connected nodes.
+			resp.Online = api.h.nodeNotifier.IsConnected(node.MachineKey)
+
+			response[index] = resp
 		}
 
 		return &v1.ListNodesResponse{Nodes: response}, nil
@@ -340,13 +358,18 @@ func (api headscaleV1APIServer) ListNodes(
 
 	response := make([]*v1.Node, len(nodes))
 	for index, node := range nodes {
-		m := node.Proto()
+		resp := node.Proto()
+
+		// Populate the online field based on
+		// currently connected nodes.
+		resp.Online = api.h.nodeNotifier.IsConnected(node.MachineKey)
+
 		validTags, invalidTags := api.h.ACLPolicy.TagsOfNode(
 			&node,
 		)
-		m.InvalidTags = invalidTags
+		resp.InvalidTags = invalidTags
-		m.ValidTags = validTags
+		resp.ValidTags = validTags
-		response[index] = m
+		response[index] = resp
 	}
 
 	return &v1.ListNodesResponse{Nodes: response}, nil
@@ -521,13 +544,22 @@ func (api headscaleV1APIServer) DebugCreateNode(
 		Hostname:    "DebugTestNode",
 	}
 
-	givenName, err := api.h.db.GenerateGivenName(request.GetKey(), request.GetName())
+	var mkey key.MachinePublic
+	err = mkey.UnmarshalText([]byte(request.GetKey()))
 	if err != nil {
 		return nil, err
 	}
 
+	givenName, err := api.h.db.GenerateGivenName(mkey, request.GetName())
+	if err != nil {
+		return nil, err
+	}
+
+	nodeKey := key.NewNode()
+
 	newNode := types.Node{
-		MachineKey: request.GetKey(),
+		MachineKey: mkey,
+		NodeKey:    nodeKey.Public(),
 		Hostname:   request.GetName(),
 		GivenName:  givenName,
 		User:       *user,
@@ -535,17 +567,15 @@ func (api headscaleV1APIServer) DebugCreateNode(
 		Expiry:   &time.Time{},
 		LastSeen: &time.Time{},
 
-		HostInfo: types.HostInfo(hostinfo),
+		Hostinfo: &hostinfo,
 	}
 
-	nodeKey := key.NodePublic{}
+	log.Debug().
-	err = nodeKey.UnmarshalText([]byte(request.GetKey()))
+		Str("machine_key", mkey.ShortString()).
-	if err != nil {
+		Msg("adding debug machine via CLI, appending to registration cache")
-		log.Panic().Msg("can not add node for debug. invalid node key")
-	}
 
 	api.h.registrationCache.Set(
-		util.NodePublicKeyStripPrefix(nodeKey),
+		mkey.String(),
 		newNode,
 		registerCacheExpiration,
 	)

hscontrol/handler_legacy.go

@@ -1,15 +0,0 @@
-//go:build ts2019
-
-package hscontrol
-
-import (
-	"net/http"
-
-	"github.com/gorilla/mux"
-)
-
-func (h *Headscale) addLegacyHandlers(router *mux.Router) {
-	router.HandleFunc("/machine/{mkey}/map", h.PollNetMapHandler).
-		Methods(http.MethodPost)
-	router.HandleFunc("/machine/{mkey}", h.RegistrationHandler).Methods(http.MethodPost)
-}

hscontrol/handler_placeholder.go

@@ -1,8 +0,0 @@
-//go:build !ts2019
-
-package hscontrol
-
-import "github.com/gorilla/mux"
-
-func (h *Headscale) addLegacyHandlers(router *mux.Router) {
-}

hscontrol/handlers.go

@@ -11,7 +11,6 @@ import (
 	"time"
 
 	"github.com/gorilla/mux"
-	"github.com/juanfont/headscale/hscontrol/util"
 	"github.com/rs/zerolog/log"
 	"tailscale.com/tailcfg"
 	"tailscale.com/types/key"
@@ -63,26 +62,6 @@ func (h *Headscale) KeyHandler(
 	// New Tailscale clients send a 'v' parameter to indicate the CurrentCapabilityVersion
 	capVer, err := parseCabailityVersion(req)
 	if err != nil {
-		if errors.Is(err, ErrNoCapabilityVersion) {
-			log.Debug().
-				Str("handler", "/key").
-				Msg("New legacy client")
-			// Old clients don't send a 'v' parameter, so we send the legacy public key
-			writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
-			writer.WriteHeader(http.StatusOK)
-			_, err := writer.Write(
-				[]byte(util.MachinePublicKeyStripPrefix(h.privateKey2019.Public())),
-			)
-			if err != nil {
-				log.Error().
-					Caller().
-					Err(err).
-					Msg("Failed to write response")
-			}
-
-			return
-		}
-
 		log.Error().
 			Caller().
 			Err(err).
@@ -101,7 +80,7 @@ func (h *Headscale) KeyHandler(
 
 	log.Debug().
 		Str("handler", "/key").
-		Int("v", int(capVer)).
+		Int("cap_ver", int(capVer)).
 		Msg("New noise client")
 	if err != nil {
 		writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
@@ -120,8 +99,7 @@ func (h *Headscale) KeyHandler(
 	// TS2021 (Tailscale v2 protocol) requires to have a different key
 	if capVer >= NoiseCapabilityVersion {
 		resp := tailcfg.OverTLSPublicKeyResponse{
-			LegacyPublicKey: h.privateKey2019.Public(),
+			PublicKey: h.noisePrivateKey.Public(),
-			PublicKey:       h.noisePrivateKey.Public(),
 		}
 		writer.Header().Set("Content-Type", "application/json")
 		writer.WriteHeader(http.StatusOK)
@@ -206,33 +184,16 @@ func (h *Headscale) RegisterWebAPI(
 	req *http.Request,
 ) {
 	vars := mux.Vars(req)
-	nodeKeyStr, ok := vars["nkey"]
+	machineKeyStr := vars["mkey"]
-
-	if !util.NodePublicKeyRegex.Match([]byte(nodeKeyStr)) {
-		log.Warn().Str("node_key", nodeKeyStr).Msg("Invalid node key passed to registration url")
-
-		writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
-		writer.WriteHeader(http.StatusUnauthorized)
-		_, err := writer.Write([]byte("Unauthorized"))
-		if err != nil {
-			log.Error().
-				Caller().
-				Err(err).
-				Msg("Failed to write response")
-		}
-
-		return
-	}
 
 	// We need to make sure we dont open for XSS style injections, if the parameter that
 	// is passed as a key is not parsable/validated as a NodePublic key, then fail to render
 	// the template and log an error.
-	var nodeKey key.NodePublic
+	var machineKey key.MachinePublic
-	err := nodeKey.UnmarshalText(
+	err := machineKey.UnmarshalText(
-		[]byte(util.NodePublicKeyEnsurePrefix(nodeKeyStr)),
+		[]byte(machineKeyStr),
 	)
-
+	if err != nil {
-	if !ok || nodeKeyStr == "" || err != nil {
 		log.Warn().Err(err).Msg("Failed to parse incoming nodekey")
 
 		writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
@@ -250,7 +211,7 @@ func (h *Headscale) RegisterWebAPI(
 
 	var content bytes.Buffer
 	if err := registerWebAPITemplate.Execute(&content, registerWebAPITemplateConfig{
-		Key: nodeKeyStr,
+		Key: machineKey.String(),
 	}); err != nil {
 		log.Error().
 			Str("func", "RegisterWebAPI").

hscontrol/mapper/mapper.go

@@ -8,6 +8,7 @@ import (
 	"net/url"
 	"os"
 	"path"
+	"slices"
 	"sort"
 	"strings"
 	"sync"
@@ -20,12 +21,11 @@ import (
 	"github.com/juanfont/headscale/hscontrol/util"
 	"github.com/klauspost/compress/zstd"
 	"github.com/rs/zerolog/log"
-	"github.com/samber/lo"
+	"golang.org/x/exp/maps"
 	"tailscale.com/envknob"
 	"tailscale.com/smallzstd"
 	"tailscale.com/tailcfg"
 	"tailscale.com/types/dnstype"
-	"tailscale.com/types/key"
 )
 
 const (
@@ -46,12 +46,9 @@ var debugDumpMapResponsePath = envknob.String("HEADSCALE_DEBUG_DUMP_MAPRESPONSE_
 // - Keep information about the previous mapresponse so we can send a diff
 // - Store hashes
 // - Create a "minifier" that removes info not needed for the node
+// - some sort of batching, wait for 5 or 60 seconds before sending
 
 type Mapper struct {
-	privateKey2019 *key.MachinePrivate
-	isNoise        bool
-	capVer         tailcfg.CapabilityVersion
-
 	// Configuration
 	// TODO(kradalby): figure out if this is the format we want this in
 	derpMap          *tailcfg.DERPMap
@@ -66,16 +63,19 @@ type Mapper struct {
 
 	// Map isnt concurrency safe, so we need to ensure
 	// only one func is accessing it over time.
-	mu    sync.Mutex
+	mu      sync.Mutex
-	peers map[uint64]*types.Node
+	peers   map[uint64]*types.Node
+	patches map[uint64][]patch
+}
+
+type patch struct {
+	timestamp time.Time
+	change    *tailcfg.PeerChange
 }
 
 func NewMapper(
 	node *types.Node,
 	peers types.Nodes,
-	privateKey *key.MachinePrivate,
-	isNoise bool,
-	capVer tailcfg.CapabilityVersion,
 	derpMap *tailcfg.DERPMap,
 	baseDomain string,
 	dnsCfg *tailcfg.DNSConfig,
@@ -84,17 +84,12 @@ func NewMapper(
 ) *Mapper {
 	log.Debug().
 		Caller().
-		Bool("noise", isNoise).
 		Str("node", node.Hostname).
 		Msg("creating new mapper")
 
 	uid, _ := util.GenerateRandomStringDNSSafe(mapperIDLength)
 
 	return &Mapper{
-		privateKey2019: privateKey,
-		isNoise:        isNoise,
-		capVer:         capVer,
-
 		derpMap:          derpMap,
 		baseDomain:       baseDomain,
 		dnsCfg:           dnsCfg,
@@ -106,7 +101,8 @@ func NewMapper(
 		seq:     0,
 
 		// TODO: populate
-		peers: peers.IDMap(),
+		peers:   peers.IDMap(),
+		patches: make(map[uint64][]patch),
 	}
 }
 
@@ -195,7 +191,7 @@ func addNextDNSMetadata(resolvers []*dnstype.Resolver, node *types.Node) {
 		if strings.HasPrefix(resolver.Addr, nextDNSDoHPrefix) {
 			attrs := url.Values{
 				"device_name":  []string{node.Hostname},
-				"device_model": []string{node.HostInfo.OS},
+				"device_model": []string{node.Hostinfo.OS},
 			}
 
 			if len(node.IPAddresses) > 0 {
@@ -212,10 +208,11 @@ func addNextDNSMetadata(resolvers []*dnstype.Resolver, node *types.Node) {
 func (m *Mapper) fullMapResponse(
 	node *types.Node,
 	pol *policy.ACLPolicy,
+	capVer tailcfg.CapabilityVersion,
 ) (*tailcfg.MapResponse, error) {
 	peers := nodeMapToList(m.peers)
 
-	resp, err := m.baseWithConfigMapResponse(node, pol)
+	resp, err := m.baseWithConfigMapResponse(node, pol, capVer)
 	if err != nil {
 		return nil, err
 	}
@@ -224,7 +221,7 @@ func (m *Mapper) fullMapResponse(
 		resp,
 		pol,
 		node,
-		m.capVer,
+		capVer,
 		peers,
 		peers,
 		m.baseDomain,
@@ -247,13 +244,22 @@ func (m *Mapper) FullMapResponse(
 	m.mu.Lock()
 	defer m.mu.Unlock()
 
-	resp, err := m.fullMapResponse(node, pol)
+	peers := maps.Keys(m.peers)
-	if err != nil {
+	peersWithPatches := maps.Keys(m.patches)
-		return nil, err
+	slices.Sort(peers)
+	slices.Sort(peersWithPatches)
+
+	if len(peersWithPatches) > 0 {
+		log.Debug().
+			Str("node", node.Hostname).
+			Uints64("peers", peers).
+			Uints64("pending_patches", peersWithPatches).
+			Msgf("node requested full map response, but has pending patches")
 	}
 
-	if m.isNoise {
+	resp, err := m.fullMapResponse(node, pol, mapRequest.Version)
-		return m.marshalMapResponse(mapRequest, resp, node, mapRequest.Compress)
+	if err != nil {
+		return nil, err
 	}
 
 	return m.marshalMapResponse(mapRequest, resp, node, mapRequest.Compress)
@@ -267,15 +273,11 @@ func (m *Mapper) LiteMapResponse(
 	node *types.Node,
 	pol *policy.ACLPolicy,
 ) ([]byte, error) {
-	resp, err := m.baseWithConfigMapResponse(node, pol)
+	resp, err := m.baseWithConfigMapResponse(node, pol, mapRequest.Version)
 	if err != nil {
 		return nil, err
 	}
 
-	if m.isNoise {
-		return m.marshalMapResponse(mapRequest, resp, node, mapRequest.Compress)
-	}
-
 	return m.marshalMapResponse(mapRequest, resp, node, mapRequest.Compress)
 }
 
@@ -292,10 +294,12 @@ func (m *Mapper) KeepAliveResponse(
 func (m *Mapper) DERPMapResponse(
 	mapRequest tailcfg.MapRequest,
 	node *types.Node,
-	derpMap tailcfg.DERPMap,
+	derpMap *tailcfg.DERPMap,
 ) ([]byte, error) {
+	m.derpMap = derpMap
+
 	resp := m.baseMapResponse()
-	resp.DERPMap = &derpMap
+	resp.DERPMap = derpMap
 
 	return m.marshalMapResponse(mapRequest, &resp, node, mapRequest.Compress)
 }
@@ -305,18 +309,29 @@ func (m *Mapper) PeerChangedResponse(
 	node *types.Node,
 	changed types.Nodes,
 	pol *policy.ACLPolicy,
+	messages ...string,
 ) ([]byte, error) {
 	m.mu.Lock()
 	defer m.mu.Unlock()
 
-	lastSeen := make(map[tailcfg.NodeID]bool)
-
 	// Update our internal map.
 	for _, node := range changed {
-		m.peers[node.ID] = node
+		if patches, ok := m.patches[node.ID]; ok {
+			// preserve online status in case the patch has an outdated one
+			online := node.IsOnline
 
-		// We have just seen the node, let the peers update their list.
+			for _, p := range patches {
-		lastSeen[tailcfg.NodeID(node.ID)] = true
+				// TODO(kradalby): Figure if this needs to be sorted by timestamp
+				node.ApplyPeerChange(p.change)
+			}
+
+			// Ensure the patches are not applied again later
+			delete(m.patches, node.ID)
+
+			node.IsOnline = online
+		}
+
+		m.peers[node.ID] = node
 	}
 
 	resp := m.baseMapResponse()
@@ -325,7 +340,7 @@ func (m *Mapper) PeerChangedResponse(
 		&resp,
 		pol,
 		node,
-		m.capVer,
+		mapRequest.Version,
 		nodeMapToList(m.peers),
 		changed,
 		m.baseDomain,
@@ -336,11 +351,55 @@ func (m *Mapper) PeerChangedResponse(
 		return nil, err
 	}
 
-	// resp.PeerSeenChange = lastSeen
+	return m.marshalMapResponse(mapRequest, &resp, node, mapRequest.Compress, messages...)
+}
+
+// PeerChangedPatchResponse creates a patch MapResponse with
+// incoming update from a state change.
+func (m *Mapper) PeerChangedPatchResponse(
+	mapRequest tailcfg.MapRequest,
+	node *types.Node,
+	changed []*tailcfg.PeerChange,
+	pol *policy.ACLPolicy,
+) ([]byte, error) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	sendUpdate := false
+	// patch the internal map
+	for _, change := range changed {
+		if peer, ok := m.peers[uint64(change.NodeID)]; ok {
+			peer.ApplyPeerChange(change)
+			sendUpdate = true
+		} else {
+			log.Trace().Str("node", node.Hostname).Msgf("Node with ID %s is missing from mapper for Node %s, saving patch for when node is available", change.NodeID, node.Hostname)
+
+			p := patch{
+				timestamp: time.Now(),
+				change:    change,
+			}
+
+			if patches, ok := m.patches[uint64(change.NodeID)]; ok {
+				patches := append(patches, p)
+
+				m.patches[uint64(change.NodeID)] = patches
+			} else {
+				m.patches[uint64(change.NodeID)] = []patch{p}
+			}
+		}
+	}
+
+	if !sendUpdate {
+		return nil, nil
+	}
+
+	resp := m.baseMapResponse()
+	resp.PeersChangedPatch = changed
 
 	return m.marshalMapResponse(mapRequest, &resp, node, mapRequest.Compress)
 }
 
+// TODO(kradalby): We need some integration tests for this.
 func (m *Mapper) PeerRemovedResponse(
 	mapRequest tailcfg.MapRequest,
 	node *types.Node,
@@ -349,13 +408,23 @@ func (m *Mapper) PeerRemovedResponse(
 	m.mu.Lock()
 	defer m.mu.Unlock()
 
+	// Some nodes might have been removed already
+	// so we dont want to ask downstream to remove
+	// twice, than can cause a panic in tailscaled.
+	notYetRemoved := []tailcfg.NodeID{}
+
 	// remove from our internal map
 	for _, id := range removed {
+		if _, ok := m.peers[uint64(id)]; ok {
+			notYetRemoved = append(notYetRemoved, id)
+		}
+
 		delete(m.peers, uint64(id))
+		delete(m.patches, uint64(id))
 	}
 
 	resp := m.baseMapResponse()
-	resp.PeersRemoved = removed
+	resp.PeersRemoved = notYetRemoved
 
 	return m.marshalMapResponse(mapRequest, &resp, node, mapRequest.Compress)
 }
@@ -365,20 +434,10 @@ func (m *Mapper) marshalMapResponse(
 	resp *tailcfg.MapResponse,
 	node *types.Node,
 	compression string,
+	messages ...string,
 ) ([]byte, error) {
 	atomic.AddUint64(&m.seq, 1)
 
-	var machineKey key.MachinePublic
-	err := machineKey.UnmarshalText([]byte(util.MachinePublicKeyEnsurePrefix(node.MachineKey)))
-	if err != nil {
-		log.Error().
-			Caller().
-			Err(err).
-			Msg("Cannot parse client key")
-
-		return nil, err
-	}
-
 	jsonBody, err := json.Marshal(resp)
 	if err != nil {
 		log.Error().
@@ -389,11 +448,27 @@ func (m *Mapper) marshalMapResponse(
 
 	if debugDumpMapResponsePath != "" {
 		data := map[string]interface{}{
+			"Messages":    messages,
 			"MapRequest":  mapRequest,
 			"MapResponse": resp,
 		}
 
-		body, err := json.Marshal(data)
+		responseType := "keepalive"
+
+		switch {
+		case resp.Peers != nil && len(resp.Peers) > 0:
+			responseType = "full"
+		case resp.Peers == nil && resp.PeersChanged == nil && resp.PeersChangedPatch == nil:
+			responseType = "lite"
+		case resp.PeersChanged != nil && len(resp.PeersChanged) > 0:
+			responseType = "changed"
+		case resp.PeersChangedPatch != nil && len(resp.PeersChangedPatch) > 0:
+			responseType = "patch"
+		case resp.PeersRemoved != nil && len(resp.PeersRemoved) > 0:
+			responseType = "removed"
+		}
+
+		body, err := json.MarshalIndent(data, "", "  ")
 		if err != nil {
 			log.Error().
 				Caller().
@@ -412,7 +487,7 @@ func (m *Mapper) marshalMapResponse(
 
 		mapResponsePath := path.Join(
 			mPath,
-			fmt.Sprintf("%d-%s-%d.json", now, m.uid, atomic.LoadUint64(&m.seq)),
+			fmt.Sprintf("%d-%s-%d-%s.json", now, m.uid, atomic.LoadUint64(&m.seq), responseType),
 		)
 
 		log.Trace().Msgf("Writing MapResponse to %s", mapResponsePath)
@@ -425,15 +500,8 @@ func (m *Mapper) marshalMapResponse(
 	var respBody []byte
 	if compression == util.ZstdCompression {
 		respBody = zstdEncode(jsonBody)
-		if !m.isNoise { // if legacy protocol
-			respBody = m.privateKey2019.SealTo(machineKey, respBody)
-		}
 	} else {
-		if !m.isNoise { // if legacy protocol
+		respBody = jsonBody
-			respBody = m.privateKey2019.SealTo(machineKey, jsonBody)
-		} else {
-			respBody = jsonBody
-		}
 	}
 
 	data := make([]byte, reservedResponseHeaderSize)
@@ -443,32 +511,6 @@ func (m *Mapper) marshalMapResponse(
 	return data, nil
 }
 
-// MarshalResponse takes an Tailscale Response, marhsal it to JSON.
-// If isNoise is set, then the JSON body will be returned
-// If !isNoise and privateKey2019 is set, the JSON body will be sealed in a Nacl box.
-func MarshalResponse(
-	resp interface{},
-	isNoise bool,
-	privateKey2019 *key.MachinePrivate,
-	machineKey key.MachinePublic,
-) ([]byte, error) {
-	jsonBody, err := json.Marshal(resp)
-	if err != nil {
-		log.Error().
-			Caller().
-			Err(err).
-			Msg("Cannot marshal response")
-
-		return nil, err
-	}
-
-	if !isNoise && privateKey2019 != nil {
-		return privateKey2019.SealTo(machineKey, jsonBody), nil
-	}
-
-	return jsonBody, nil
-}
-
 func zstdEncode(in []byte) []byte {
 	encoder, ok := zstdEncoderPool.Get().(*zstd.Encoder)
 	if !ok {
@@ -502,6 +544,7 @@ func (m *Mapper) baseMapResponse() tailcfg.MapResponse {
 	resp := tailcfg.MapResponse{
 		KeepAlive:   false,
 		ControlTime: &now,
+		// TODO(kradalby): Implement PingRequest?
 	}
 
 	return resp
@@ -514,10 +557,11 @@ func (m *Mapper) baseMapResponse() tailcfg.MapResponse {
 func (m *Mapper) baseWithConfigMapResponse(
 	node *types.Node,
 	pol *policy.ACLPolicy,
+	capVer tailcfg.CapabilityVersion,
 ) (*tailcfg.MapResponse, error) {
 	resp := m.baseMapResponse()
 
-	tailnode, err := tailNode(node, m.capVer, pol, m.dnsCfg, m.baseDomain, m.randomClientPort)
+	tailnode, err := tailNode(node, capVer, pol, m.dnsCfg, m.baseDomain, m.randomClientPort)
 	if err != nil {
 		return nil, err
 	}
@@ -550,15 +594,6 @@ func nodeMapToList(nodes map[uint64]*types.Node) types.Nodes {
 	return ret
 }
 
-func filterExpiredAndNotReady(peers types.Nodes) types.Nodes {
-	return lo.Filter(peers, func(item *types.Node, index int) bool {
-		// Filter out nodes that are expired OR
-		// nodes that has no endpoints, this typically means they have
-		// registered, but are not configured.
-		return !item.IsExpired() || len(item.Endpoints) > 0
-	})
-}
-
 // appendPeerChanges mutates a tailcfg.MapResponse with all the
 // necessary changes when peers have changed.
 func appendPeerChanges(
@@ -584,9 +619,6 @@ func appendPeerChanges(
 		return err
 	}
 
-	// Filter out peers that have expired.
-	changed = filterExpiredAndNotReady(changed)
-
 	// If there are filter rules present, see if there are any nodes that cannot
 	// access eachother at all and remove them from the peers.
 	if len(rules) > 0 {
@@ -622,8 +654,5 @@ func appendPeerChanges(
 	resp.UserProfiles = profiles
 	resp.SSHPolicy = sshPolicy
 
-	// TODO(kradalby): This currently does not take last seen in keepalives into account
-	resp.OnlineChange = peers.OnlineNodeMap()
-
 	return nil
 }

hscontrol/mapper/mapper_test.go

@@ -166,10 +166,16 @@ func Test_fullMapResponse(t *testing.T) {
 	expire := time.Date(2500, time.November, 11, 23, 0, 0, 0, time.UTC)
 
 	mini := &types.Node{
-		ID:          0,
+		ID: 0,
-		MachineKey:  "mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
+		MachineKey: mustMK(
-		NodeKey:     "nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+			"mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
-		DiscoKey:    "discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+		),
+		NodeKey: mustNK(
+			"nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+		),
+		DiscoKey: mustDK(
+			"discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+		),
 		IPAddresses: []netip.Addr{netip.MustParseAddr("100.64.0.1")},
 		Hostname:    "mini",
 		GivenName:   "mini",
@@ -180,8 +186,7 @@ func Test_fullMapResponse(t *testing.T) {
 		AuthKey:     &types.PreAuthKey{},
 		LastSeen:    &lastSeen,
 		Expiry:      &expire,
-		HostInfo:    types.HostInfo{},
+		Hostinfo:    &tailcfg.Hostinfo{},
-		Endpoints:   []string{},
 		Routes: []types.Route{
 			{
 				Prefix:     types.IPPrefix(netip.MustParsePrefix("0.0.0.0/0")),
@@ -226,14 +231,12 @@ func Test_fullMapResponse(t *testing.T) {
 			netip.MustParsePrefix("0.0.0.0/0"),
 			netip.MustParsePrefix("192.168.0.0/24"),
 		},
-		Endpoints:         []string{},
 		DERP:              "127.3.3.40:0",
 		Hostinfo:          hiview(tailcfg.Hostinfo{}),
 		Created:           created,
 		Tags:              []string{},
 		PrimaryRoutes:     []netip.Prefix{netip.MustParsePrefix("192.168.0.0/24")},
 		LastSeen:          &lastSeen,
-		Online:            new(bool),
 		MachineAuthorized: true,
 		Capabilities: []tailcfg.NodeCapability{
 			tailcfg.CapabilityFileSharing,
@@ -244,10 +247,16 @@ func Test_fullMapResponse(t *testing.T) {
 	}
 
 	peer1 := &types.Node{
-		ID:          1,
+		ID: 1,
-		MachineKey:  "mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
+		MachineKey: mustMK(
-		NodeKey:     "nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+			"mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
-		DiscoKey:    "discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+		),
+		NodeKey: mustNK(
+			"nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+		),
+		DiscoKey: mustDK(
+			"discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+		),
 		IPAddresses: []netip.Addr{netip.MustParseAddr("100.64.0.2")},
 		Hostname:    "peer1",
 		GivenName:   "peer1",
@@ -256,8 +265,7 @@ func Test_fullMapResponse(t *testing.T) {
 		ForcedTags:  []string{},
 		LastSeen:    &lastSeen,
 		Expiry:      &expire,
-		HostInfo:    types.HostInfo{},
+		Hostinfo:    &tailcfg.Hostinfo{},
-		Endpoints:   []string{},
 		Routes:      []types.Route{},
 		CreatedAt:   created,
 	}
@@ -278,14 +286,12 @@ func Test_fullMapResponse(t *testing.T) {
 		),
 		Addresses:         []netip.Prefix{netip.MustParsePrefix("100.64.0.2/32")},
 		AllowedIPs:        []netip.Prefix{netip.MustParsePrefix("100.64.0.2/32")},
-		Endpoints:         []string{},
 		DERP:              "127.3.3.40:0",
 		Hostinfo:          hiview(tailcfg.Hostinfo{}),
 		Created:           created,
 		Tags:              []string{},
 		PrimaryRoutes:     []netip.Prefix{},
 		LastSeen:          &lastSeen,
-		Online:            new(bool),
 		MachineAuthorized: true,
 		Capabilities: []tailcfg.NodeCapability{
 			tailcfg.CapabilityFileSharing,
@@ -296,10 +302,16 @@ func Test_fullMapResponse(t *testing.T) {
 	}
 
 	peer2 := &types.Node{
-		ID:          2,
+		ID: 2,
-		MachineKey:  "mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
+		MachineKey: mustMK(
-		NodeKey:     "nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+			"mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
-		DiscoKey:    "discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+		),
+		NodeKey: mustNK(
+			"nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+		),
+		DiscoKey: mustDK(
+			"discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+		),
 		IPAddresses: []netip.Addr{netip.MustParseAddr("100.64.0.3")},
 		Hostname:    "peer2",
 		GivenName:   "peer2",
@@ -308,8 +320,7 @@ func Test_fullMapResponse(t *testing.T) {
 		ForcedTags:  []string{},
 		LastSeen:    &lastSeen,
 		Expiry:      &expire,
-		HostInfo:    types.HostInfo{},
+		Hostinfo:    &tailcfg.Hostinfo{},
-		Endpoints:   []string{},
 		Routes:      []types.Route{},
 		CreatedAt:   created,
 	}
@@ -387,7 +398,6 @@ func Test_fullMapResponse(t *testing.T) {
 				DNSConfig:       &tailcfg.DNSConfig{},
 				Domain:          "",
 				CollectServices: "false",
-				OnlineChange:    map[tailcfg.NodeID]bool{tailPeer1.ID: false},
 				PacketFilter:    []tailcfg.FilterRule{},
 				UserProfiles:    []tailcfg.UserProfile{{LoginName: "mini", DisplayName: "mini"}},
 				SSHPolicy:       &tailcfg.SSHPolicy{Rules: []*tailcfg.SSHRule{}},
@@ -429,10 +439,6 @@ func Test_fullMapResponse(t *testing.T) {
 				DNSConfig:       &tailcfg.DNSConfig{},
 				Domain:          "",
 				CollectServices: "false",
-				OnlineChange: map[tailcfg.NodeID]bool{
-					tailPeer1.ID:             false,
-					tailcfg.NodeID(peer2.ID): false,
-				},
 				PacketFilter: []tailcfg.FilterRule{
 					{
 						SrcIPs: []string{"100.64.0.2/32"},
@@ -459,9 +465,6 @@ func Test_fullMapResponse(t *testing.T) {
 			mappy := NewMapper(
 				tt.node,
 				tt.peers,
-				nil,
-				false,
-				0,
 				tt.derpMap,
 				tt.baseDomain,
 				tt.dnsConfig,
@@ -472,6 +475,7 @@ func Test_fullMapResponse(t *testing.T) {
 			got, err := mappy.fullMapResponse(
 				tt.node,
 				tt.pol,
+				0,
 			)
 
 			if (err != nil) != tt.wantErr {

hscontrol/mapper/tail.go

@@ -52,21 +52,6 @@ func tailNode(
 	baseDomain string,
 	randomClientPort bool,
 ) (*tailcfg.Node, error) {
-	nodeKey, err := node.NodePublicKey()
-	if err != nil {
-		return nil, err
-	}
-
-	machineKey, err := node.MachinePublicKey()
-	if err != nil {
-		return nil, err
-	}
-
-	discoKey, err := node.DiscoPublicKey()
-	if err != nil {
-		return nil, err
-	}
-
 	addrs := node.IPAddresses.Prefixes()
 
 	allowedIPs := append(
@@ -87,8 +72,8 @@ func tailNode(
 	}
 
 	var derp string
-	if node.HostInfo.NetInfo != nil {
+	if node.Hostinfo.NetInfo != nil {
-		derp = fmt.Sprintf("127.3.3.40:%d", node.HostInfo.NetInfo.PreferredDERP)
+		derp = fmt.Sprintf("127.3.3.40:%d", node.Hostinfo.NetInfo.PreferredDERP)
 	} else {
 		derp = "127.3.3.40:0" // Zero means disconnected or unknown.
 	}
@@ -102,13 +87,9 @@ func tailNode(
 
 	hostname, err := node.GetFQDN(dnsConfig, baseDomain)
 	if err != nil {
-		return nil, err
+		return nil, fmt.Errorf("tailNode, failed to create FQDN: %s", err)
 	}
 
-	hostInfo := node.GetHostInfo()
-
-	online := node.IsOnline()
-
 	tags, _ := pol.TagsOfNode(node)
 	tags = lo.Uniq(append(tags, node.ForcedTags...))
 
@@ -118,28 +99,30 @@ func tailNode(
 			strconv.FormatUint(node.ID, util.Base10),
 		), // in headscale, unlike tailcontrol server, IDs are permanent
 		Name: hostname,
+		Cap:  capVer,
 
 		User: tailcfg.UserID(node.UserID),
 
-		Key:       nodeKey,
+		Key:       node.NodeKey,
 		KeyExpiry: keyExpiry,
 
-		Machine:    machineKey,
+		Machine:    node.MachineKey,
-		DiscoKey:   discoKey,
+		DiscoKey:   node.DiscoKey,
 		Addresses:  addrs,
 		AllowedIPs: allowedIPs,
 		Endpoints:  node.Endpoints,
 		DERP:       derp,
-		Hostinfo:   hostInfo.View(),
+		Hostinfo:   node.Hostinfo.View(),
 		Created:    node.CreatedAt,
 
+		Online: node.IsOnline,
+
 		Tags: tags,
 
 		PrimaryRoutes: primaryPrefixes,
 
-		LastSeen:          node.LastSeen,
-		Online:            &online,
 		MachineAuthorized: !node.IsExpired(),
+		Expired:           node.IsExpired(),
 	}
 
 	//   - 74: 2023-09-18: Client understands NodeCapMap
@@ -170,5 +153,11 @@ func tailNode(
 		tNode.Capabilities = append(tNode.Capabilities, tailcfg.NodeAttrDisableUPnP)
 	}
 
+	if node.IsOnline == nil || !*node.IsOnline {
+		// LastSeen is only set when node is
+		// not connected to the control server.
+		tNode.LastSeen = node.LastSeen
+	}
+
 	return &tNode, nil
 }

hscontrol/mapper/tail_test.go

@@ -53,21 +53,42 @@ func TestTailNode(t *testing.T) {
 		wantErr    bool
 	}{
 		{
-			name:       "empty-node",
+			name: "empty-node",
-			node:       &types.Node{},
+			node: &types.Node{
+				Hostinfo: &tailcfg.Hostinfo{},
+			},
 			pol:        &policy.ACLPolicy{},
 			dnsConfig:  &tailcfg.DNSConfig{},
 			baseDomain: "",
-			want:       nil,
+			want: &tailcfg.Node{
-			wantErr:    true,
+				StableID:          "0",
+				Addresses:         []netip.Prefix{},
+				AllowedIPs:        []netip.Prefix{},
+				DERP:              "127.3.3.40:0",
+				Hostinfo:          hiview(tailcfg.Hostinfo{}),
+				Tags:              []string{},
+				PrimaryRoutes:     []netip.Prefix{},
+				MachineAuthorized: true,
+				Capabilities: []tailcfg.NodeCapability{
+					"https://tailscale.com/cap/file-sharing", "https://tailscale.com/cap/is-admin",
+					"https://tailscale.com/cap/ssh", "debug-disable-upnp",
+				},
+			},
+			wantErr: false,
 		},
 		{
 			name: "minimal-node",
 			node: &types.Node{
-				ID:         0,
+				ID: 0,
-				MachineKey: "mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
+				MachineKey: mustMK(
-				NodeKey:    "nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+					"mkey:f08305b4ee4250b95a70f3b7504d048d75d899993c624a26d422c67af0422507",
-				DiscoKey:   "discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+				),
+				NodeKey: mustNK(
+					"nodekey:9b2ffa7e08cc421a3d2cca9012280f6a236fd0de0b4ce005b30a98ad930306fe",
+				),
+				DiscoKey: mustDK(
+					"discokey:cf7b0fd05da556fdc3bab365787b506fd82d64a70745db70e00e86c1b1c03084",
+				),
 				IPAddresses: []netip.Addr{
 					netip.MustParseAddr("100.64.0.1"),
 				},
@@ -82,8 +103,7 @@ func TestTailNode(t *testing.T) {
 				AuthKey:    &types.PreAuthKey{},
 				LastSeen:   &lastSeen,
 				Expiry:     &expire,
-				HostInfo:   types.HostInfo{},
+				Hostinfo:   &tailcfg.Hostinfo{},
-				Endpoints:  []string{},
 				Routes: []types.Route{
 					{
 						Prefix:     types.IPPrefix(netip.MustParsePrefix("0.0.0.0/0")),
@@ -133,10 +153,9 @@ func TestTailNode(t *testing.T) {
 					netip.MustParsePrefix("0.0.0.0/0"),
 					netip.MustParsePrefix("192.168.0.0/24"),
 				},
-				Endpoints: []string{},
+				DERP:     "127.3.3.40:0",
-				DERP:      "127.3.3.40:0",
+				Hostinfo: hiview(tailcfg.Hostinfo{}),
-				Hostinfo:  hiview(tailcfg.Hostinfo{}),
+				Created:  created,
-				Created:   created,
 
 				Tags: []string{},
 
@@ -145,7 +164,6 @@ func TestTailNode(t *testing.T) {
 				},
 
 				LastSeen:          &lastSeen,
-				Online:            new(bool),
 				MachineAuthorized: true,
 
 				Capabilities: []tailcfg.NodeCapability{

hscontrol/notifier/notifier.go

@@ -1,11 +1,14 @@
 package notifier
 
 import (
+	"fmt"
+	"strings"
 	"sync"
 
 	"github.com/juanfont/headscale/hscontrol/types"
 	"github.com/juanfont/headscale/hscontrol/util"
 	"github.com/rs/zerolog/log"
+	"tailscale.com/types/key"
 )
 
 type Notifier struct {
@@ -17,9 +20,9 @@ func NewNotifier() *Notifier {
 	return &Notifier{}
 }
 
-func (n *Notifier) AddNode(machineKey string, c chan<- types.StateUpdate) {
+func (n *Notifier) AddNode(machineKey key.MachinePublic, c chan<- types.StateUpdate) {
-	log.Trace().Caller().Str("key", machineKey).Msg("acquiring lock to add node")
+	log.Trace().Caller().Str("key", machineKey.ShortString()).Msg("acquiring lock to add node")
-	defer log.Trace().Caller().Str("key", machineKey).Msg("releasing lock to add node")
+	defer log.Trace().Caller().Str("key", machineKey.ShortString()).Msg("releasing lock to add node")
 
 	n.l.Lock()
 	defer n.l.Unlock()
@@ -28,17 +31,17 @@ func (n *Notifier) AddNode(machineKey string, c chan<- types.StateUpdate) {
 		n.nodes = make(map[string]chan<- types.StateUpdate)
 	}
 
-	n.nodes[machineKey] = c
+	n.nodes[machineKey.String()] = c
 
 	log.Trace().
-		Str("machine_key", machineKey).
+		Str("machine_key", machineKey.ShortString()).
 		Int("open_chans", len(n.nodes)).
 		Msg("Added new channel")
 }
 
-func (n *Notifier) RemoveNode(machineKey string) {
+func (n *Notifier) RemoveNode(machineKey key.MachinePublic) {
-	log.Trace().Caller().Str("key", machineKey).Msg("acquiring lock to remove node")
+	log.Trace().Caller().Str("key", machineKey.ShortString()).Msg("acquiring lock to remove node")
-	defer log.Trace().Caller().Str("key", machineKey).Msg("releasing lock to remove node")
+	defer log.Trace().Caller().Str("key", machineKey.ShortString()).Msg("releasing lock to remove node")
 
 	n.l.Lock()
 	defer n.l.Unlock()
@@ -47,14 +50,27 @@ func (n *Notifier) RemoveNode(machineKey string) {
 		return
 	}
 
-	delete(n.nodes, machineKey)
+	delete(n.nodes, machineKey.String())
 
 	log.Trace().
-		Str("machine_key", machineKey).
+		Str("machine_key", machineKey.ShortString()).
 		Int("open_chans", len(n.nodes)).
 		Msg("Removed channel")
 }
 
+// IsConnected reports if a node is connected to headscale and has a
+// poll session open.
+func (n *Notifier) IsConnected(machineKey key.MachinePublic) bool {
+	n.l.RLock()
+	defer n.l.RUnlock()
+
+	if _, ok := n.nodes[machineKey.String()]; ok {
+		return true
+	}
+
+	return false
+}
+
 func (n *Notifier) NotifyAll(update types.StateUpdate) {
 	n.NotifyWithIgnore(update)
 }
@@ -78,3 +94,31 @@ func (n *Notifier) NotifyWithIgnore(update types.StateUpdate, ignore ...string)
 		c <- update
 	}
 }
+
+func (n *Notifier) NotifyByMachineKey(update types.StateUpdate, mKey key.MachinePublic) {
+	log.Trace().Caller().Interface("type", update.Type).Msg("acquiring lock to notify")
+	defer log.Trace().
+		Caller().
+		Interface("type", update.Type).
+		Msg("releasing lock, finished notifing")
+
+	n.l.RLock()
+	defer n.l.RUnlock()
+
+	if c, ok := n.nodes[mKey.String()]; ok {
+		c <- update
+	}
+}
+
+func (n *Notifier) String() string {
+	n.l.RLock()
+	defer n.l.RUnlock()
+
+	str := []string{"Notifier, in map:\n"}
+
+	for k, v := range n.nodes {
+		str = append(str, fmt.Sprintf("\t%s: %v\n", k, v))
+	}
+
+	return strings.Join(str, "")
+}

hscontrol/oidc.go

@@ -124,42 +124,28 @@ func (h *Headscale) determineTokenExpiration(idTokenExpiration time.Time) time.T
 
 // RegisterOIDC redirects to the OIDC provider for authentication
 // Puts NodeKey in cache so the callback can retrieve it using the oidc state param
-// Listens in /oidc/register/:nKey.
+// Listens in /oidc/register/:mKey.
 func (h *Headscale) RegisterOIDC(
 	writer http.ResponseWriter,
 	req *http.Request,
 ) {
 	vars := mux.Vars(req)
-	nodeKeyStr, ok := vars["nkey"]
+	machineKeyStr, ok := vars["mkey"]
 
 	log.Debug().
 		Caller().
-		Str("node_key", nodeKeyStr).
+		Str("machine_key", machineKeyStr).
 		Bool("ok", ok).
 		Msg("Received oidc register call")
 
-	if !util.NodePublicKeyRegex.Match([]byte(nodeKeyStr)) {
-		log.Warn().Str("node_key", nodeKeyStr).Msg("Invalid node key passed to registration url")
-
-		writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
-		writer.WriteHeader(http.StatusUnauthorized)
-		_, err := writer.Write([]byte("Unauthorized"))
-		if err != nil {
-			util.LogErr(err, "Failed to write response")
-		}
-
-		return
-	}
-
 	// We need to make sure we dont open for XSS style injections, if the parameter that
 	// is passed as a key is not parsable/validated as a NodePublic key, then fail to render
 	// the template and log an error.
-	var nodeKey key.NodePublic
+	var machineKey key.MachinePublic
-	err := nodeKey.UnmarshalText(
+	err := machineKey.UnmarshalText(
-		[]byte(util.NodePublicKeyEnsurePrefix(nodeKeyStr)),
+		[]byte(machineKeyStr),
 	)
-
+	if err != nil {
-	if !ok || nodeKeyStr == "" || err != nil {
 		log.Warn().
 			Err(err).
 			Msg("Failed to parse incoming nodekey in OIDC registration")
@@ -188,7 +174,7 @@ func (h *Headscale) RegisterOIDC(
 	// place the node key into the state cache, so it can be retrieved later
 	h.registrationCache.Set(
 		stateStr,
-		util.NodePublicKeyStripPrefix(nodeKey),
+		machineKey,
 		registerCacheExpiration,
 	)
 
@@ -266,7 +252,7 @@ func (h *Headscale) OIDCCallback(
 		return
 	}
 
-	nodeKey, nodeExists, err := h.validateNodeForOIDCCallback(
+	machineKey, nodeExists, err := h.validateNodeForOIDCCallback(
 		writer,
 		state,
 		claims,
@@ -294,7 +280,7 @@ func (h *Headscale) OIDCCallback(
 		return
 	}
 
-	if err := h.registerNodeForOIDCCallback(writer, user, nodeKey, idTokenExpiry); err != nil {
+	if err := h.registerNodeForOIDCCallback(writer, user, machineKey, idTokenExpiry); err != nil {
 		return
 	}
 
@@ -539,10 +525,10 @@ func (h *Headscale) validateNodeForOIDCCallback(
 	state string,
 	claims *IDTokenClaims,
 	expiry time.Time,
-) (*key.NodePublic, bool, error) {
+) (*key.MachinePublic, bool, error) {
 	// retrieve nodekey from state cache
-	nodeKeyIf, nodeKeyFound := h.registrationCache.Get(state)
+	machineKeyIf, machineKeyFound := h.registrationCache.Get(state)
-	if !nodeKeyFound {
+	if !machineKeyFound {
 		log.Trace().
 			Msg("requested node state key expired before authorisation completed")
 		writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
@@ -555,11 +541,12 @@ func (h *Headscale) validateNodeForOIDCCallback(
 		return nil, false, errOIDCNodeKeyMissing
 	}
 
-	var nodeKey key.NodePublic
+	var machineKey key.MachinePublic
-	nodeKeyFromCache, nodeKeyOK := nodeKeyIf.(string)
+	machineKey, machineKeyOK := machineKeyIf.(key.MachinePublic)
-	if !nodeKeyOK {
+	if !machineKeyOK {
 		log.Trace().
-			Msg("requested node state key is not a string")
+			Interface("got", machineKeyIf).
+			Msg("requested node state key is not a nodekey")
 		writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
 		writer.WriteHeader(http.StatusBadRequest)
 		_, err := writer.Write([]byte("state is invalid"))
@@ -570,29 +557,11 @@ func (h *Headscale) validateNodeForOIDCCallback(
 		return nil, false, errOIDCInvalidNodeState
 	}
 
-	err := nodeKey.UnmarshalText(
-		[]byte(util.NodePublicKeyEnsurePrefix(nodeKeyFromCache)),
-	)
-	if err != nil {
-		log.Error().
-			Str("nodeKey", nodeKeyFromCache).
-			Bool("nodeKeyOK", nodeKeyOK).
-			Msg("could not parse node public key")
-		writer.Header().Set("Content-Type", "text/plain; charset=utf-8")
-		writer.WriteHeader(http.StatusBadRequest)
-		_, werr := writer.Write([]byte("could not parse node public key"))
-		if werr != nil {
-			util.LogErr(err, "Failed to write response")
-		}
-
-		return nil, false, err
-	}
-
 	// retrieve node information if it exist
 	// The error is not important, because if it does not
 	// exist, then this is a new node and we will move
 	// on to registration.
-	node, _ := h.db.GetNodeByNodeKey(nodeKey)
+	node, _ := h.db.GetNodeByMachineKey(machineKey)
 
 	if node != nil {
 		log.Trace().
@@ -657,7 +626,7 @@ func (h *Headscale) validateNodeForOIDCCallback(
 		return nil, true, nil
 	}
 
-	return &nodeKey, false, nil
+	return &machineKey, false, nil
 }
 
 func getUserName(
@@ -740,13 +709,13 @@ func (h *Headscale) findOrCreateNewUserForOIDCCallback(
 func (h *Headscale) registerNodeForOIDCCallback(
 	writer http.ResponseWriter,
 	user *types.User,
-	nodeKey *key.NodePublic,
+	machineKey *key.MachinePublic,
 	expiry time.Time,
 ) error {
 	if _, err := h.db.RegisterNodeFromAuthCallback(
 		// TODO(kradalby): find a better way to use the cache across modules
 		h.registrationCache,
-		nodeKey.String(),
+		*machineKey,
 		user.Name,
 		&expiry,
 		util.RegisterMethodOIDC,

hscontrol/policy/acls.go

@@ -596,10 +596,13 @@ func excludeCorrectlyTaggedNodes(
 	}
 	// for each node if tag is in tags list, don't append it.
 	for _, node := range nodes {
-		hi := node.GetHostInfo()
-
 		found := false
-		for _, t := range hi.RequestTags {
+
+		if node.Hostinfo == nil {
+			continue
+		}
+
+		for _, t := range node.Hostinfo.RequestTags {
 			if util.StringOrPrefixListContains(tags, t) {
 				found = true
 
@@ -671,14 +674,18 @@ func expandOwnersFromTag(
 	pol *ACLPolicy,
 	tag string,
 ) ([]string, error) {
+	noTagErr := fmt.Errorf(
+		"%w. %v isn't owned by a TagOwner. Please add one first. https://tailscale.com/kb/1018/acls/#tag-owners",
+		ErrInvalidTag,
+		tag,
+	)
+	if pol == nil {
+		return []string{}, noTagErr
+	}
 	var owners []string
 	ows, ok := pol.TagOwners[tag]
 	if !ok {
-		return []string{}, fmt.Errorf(
+		return []string{}, noTagErr
-			"%w. %v isn't owned by a TagOwner. Please add one first. https://tailscale.com/kb/1018/acls/#tag-owners",
-			ErrInvalidTag,
-			tag,
-		)
 	}
 	for _, owner := range ows {
 		if isGroup(owner) {
@@ -787,8 +794,11 @@ func (pol *ACLPolicy) expandIPsFromTag(
 	for _, user := range owners {
 		nodes := filterNodesByUser(nodes, user)
 		for _, node := range nodes {
-			hi := node.GetHostInfo()
+			if node.Hostinfo == nil {
-			if util.StringOrPrefixListContains(hi.RequestTags, alias) {
+				continue
+			}
+
+			if util.StringOrPrefixListContains(node.Hostinfo.RequestTags, alias) {
 				node.IPAddresses.AppendToIPSet(&build)
 			}
 		}
@@ -882,7 +892,7 @@ func (pol *ACLPolicy) TagsOfNode(
 
 	validTagMap := make(map[string]bool)
 	invalidTagMap := make(map[string]bool)
-	for _, tag := range node.HostInfo.RequestTags {
+	for _, tag := range node.Hostinfo.RequestTags {
 		owners, err := expandOwnersFromTag(pol, tag)
 		if errors.Is(err, ErrInvalidTag) {
 			invalidTagMap[tag] = true

hscontrol/policy/acls_test.go

@@ -16,10 +16,6 @@ import (
 	"tailscale.com/tailcfg"
 )
 
-var ipComparer = cmp.Comparer(func(x, y netip.Addr) bool {
-	return x.Compare(y) == 0
-})
-
 func Test(t *testing.T) {
 	check.TestingT(t)
 }
@@ -401,6 +397,7 @@ acls:
 					User: types.User{
 						Name: "testuser",
 					},
+					Hostinfo: &tailcfg.Hostinfo{},
 				},
 			})
 
@@ -951,7 +948,7 @@ func Test_listNodesInUser(t *testing.T) {
 		t.Run(test.name, func(t *testing.T) {
 			got := filterNodesByUser(test.args.nodes, test.args.user)
 
-			if diff := cmp.Diff(test.want, got); diff != "" {
+			if diff := cmp.Diff(test.want, got, util.Comparers...); diff != "" {
 				t.Errorf("listNodesInUser() = (-want +got):\n%s", diff)
 			}
 		})
@@ -1247,7 +1244,7 @@ func Test_expandAlias(t *testing.T) {
 							netip.MustParseAddr("100.64.0.1"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:hr-webserver"},
@@ -1258,7 +1255,7 @@ func Test_expandAlias(t *testing.T) {
 							netip.MustParseAddr("100.64.0.2"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:hr-webserver"},
@@ -1388,7 +1385,7 @@ func Test_expandAlias(t *testing.T) {
 							netip.MustParseAddr("100.64.0.2"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:hr-webserver"},
@@ -1426,7 +1423,7 @@ func Test_expandAlias(t *testing.T) {
 							netip.MustParseAddr("100.64.0.1"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:accountant-webserver"},
@@ -1437,7 +1434,7 @@ func Test_expandAlias(t *testing.T) {
 							netip.MustParseAddr("100.64.0.2"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:accountant-webserver"},
@@ -1447,13 +1444,15 @@ func Test_expandAlias(t *testing.T) {
 						IPAddresses: types.NodeAddresses{
 							netip.MustParseAddr("100.64.0.3"),
 						},
-						User: types.User{Name: "marc"},
+						User:     types.User{Name: "marc"},
+						Hostinfo: &tailcfg.Hostinfo{},
 					},
 					&types.Node{
 						IPAddresses: types.NodeAddresses{
 							netip.MustParseAddr("100.64.0.4"),
 						},
-						User: types.User{Name: "joe"},
+						User:     types.User{Name: "joe"},
+						Hostinfo: &tailcfg.Hostinfo{},
 					},
 				},
 			},
@@ -1503,7 +1502,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 							netip.MustParseAddr("100.64.0.1"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:accountant-webserver"},
@@ -1514,7 +1513,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 							netip.MustParseAddr("100.64.0.2"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:accountant-webserver"},
@@ -1524,7 +1523,8 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 						IPAddresses: types.NodeAddresses{
 							netip.MustParseAddr("100.64.0.4"),
 						},
-						User: types.User{Name: "joe"},
+						User:     types.User{Name: "joe"},
+						Hostinfo: &tailcfg.Hostinfo{},
 					},
 				},
 				user: "joe",
@@ -1533,6 +1533,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 				&types.Node{
 					IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.4")},
 					User:        types.User{Name: "joe"},
+					Hostinfo:    &tailcfg.Hostinfo{},
 				},
 			},
 		},
@@ -1553,7 +1554,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 							netip.MustParseAddr("100.64.0.1"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:accountant-webserver"},
@@ -1564,7 +1565,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 							netip.MustParseAddr("100.64.0.2"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:accountant-webserver"},
@@ -1574,7 +1575,8 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 						IPAddresses: types.NodeAddresses{
 							netip.MustParseAddr("100.64.0.4"),
 						},
-						User: types.User{Name: "joe"},
+						User:     types.User{Name: "joe"},
+						Hostinfo: &tailcfg.Hostinfo{},
 					},
 				},
 				user: "joe",
@@ -1583,6 +1585,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 				&types.Node{
 					IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.4")},
 					User:        types.User{Name: "joe"},
+					Hostinfo:    &tailcfg.Hostinfo{},
 				},
 			},
 		},
@@ -1598,7 +1601,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 							netip.MustParseAddr("100.64.0.1"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "foo",
 							RequestTags: []string{"tag:accountant-webserver"},
@@ -1610,12 +1613,14 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 						},
 						User:       types.User{Name: "joe"},
 						ForcedTags: []string{"tag:accountant-webserver"},
+						Hostinfo:   &tailcfg.Hostinfo{},
 					},
 					&types.Node{
 						IPAddresses: types.NodeAddresses{
 							netip.MustParseAddr("100.64.0.4"),
 						},
-						User: types.User{Name: "joe"},
+						User:     types.User{Name: "joe"},
+						Hostinfo: &tailcfg.Hostinfo{},
 					},
 				},
 				user: "joe",
@@ -1624,6 +1629,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 				&types.Node{
 					IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.4")},
 					User:        types.User{Name: "joe"},
+					Hostinfo:    &tailcfg.Hostinfo{},
 				},
 			},
 		},
@@ -1639,7 +1645,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 							netip.MustParseAddr("100.64.0.1"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "hr-web1",
 							RequestTags: []string{"tag:hr-webserver"},
@@ -1650,7 +1656,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 							netip.MustParseAddr("100.64.0.2"),
 						},
 						User: types.User{Name: "joe"},
-						HostInfo: types.HostInfo{
+						Hostinfo: &tailcfg.Hostinfo{
 							OS:          "centos",
 							Hostname:    "hr-web2",
 							RequestTags: []string{"tag:hr-webserver"},
@@ -1660,7 +1666,8 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 						IPAddresses: types.NodeAddresses{
 							netip.MustParseAddr("100.64.0.4"),
 						},
-						User: types.User{Name: "joe"},
+						User:     types.User{Name: "joe"},
+						Hostinfo: &tailcfg.Hostinfo{},
 					},
 				},
 				user: "joe",
@@ -1671,7 +1678,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 						netip.MustParseAddr("100.64.0.1"),
 					},
 					User: types.User{Name: "joe"},
-					HostInfo: types.HostInfo{
+					Hostinfo: &tailcfg.Hostinfo{
 						OS:          "centos",
 						Hostname:    "hr-web1",
 						RequestTags: []string{"tag:hr-webserver"},
@@ -1682,7 +1689,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 						netip.MustParseAddr("100.64.0.2"),
 					},
 					User: types.User{Name: "joe"},
-					HostInfo: types.HostInfo{
+					Hostinfo: &tailcfg.Hostinfo{
 						OS:          "centos",
 						Hostname:    "hr-web2",
 						RequestTags: []string{"tag:hr-webserver"},
@@ -1692,7 +1699,8 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 					IPAddresses: types.NodeAddresses{
 						netip.MustParseAddr("100.64.0.4"),
 					},
-					User: types.User{Name: "joe"},
+					User:     types.User{Name: "joe"},
+					Hostinfo: &tailcfg.Hostinfo{},
 				},
 			},
 		},
@@ -1704,7 +1712,7 @@ func Test_excludeCorrectlyTaggedNodes(t *testing.T) {
 				test.args.nodes,
 				test.args.user,
 			)
-			if diff := cmp.Diff(test.want, got, ipComparer); diff != "" {
+			if diff := cmp.Diff(test.want, got, util.Comparers...); diff != "" {
 				t.Errorf("excludeCorrectlyTaggedNodes() (-want +got):\n%s", diff)
 			}
 		})
@@ -1935,7 +1943,7 @@ func Test_getTags(t *testing.T) {
 					User: types.User{
 						Name: "joe",
 					},
-					HostInfo: types.HostInfo{
+					Hostinfo: &tailcfg.Hostinfo{
 						RequestTags: []string{"tag:valid"},
 					},
 				},
@@ -1955,7 +1963,7 @@ func Test_getTags(t *testing.T) {
 					User: types.User{
 						Name: "joe",
 					},
-					HostInfo: types.HostInfo{
+					Hostinfo: &tailcfg.Hostinfo{
 						RequestTags: []string{"tag:valid", "tag:invalid"},
 					},
 				},
@@ -1975,7 +1983,7 @@ func Test_getTags(t *testing.T) {
 					User: types.User{
 						Name: "joe",
 					},
-					HostInfo: types.HostInfo{
+					Hostinfo: &tailcfg.Hostinfo{
 						RequestTags: []string{
 							"tag:invalid",
 							"tag:valid",
@@ -1999,7 +2007,7 @@ func Test_getTags(t *testing.T) {
 					User: types.User{
 						Name: "joe",
 					},
-					HostInfo: types.HostInfo{
+					Hostinfo: &tailcfg.Hostinfo{
 						RequestTags: []string{"tag:invalid", "very-invalid"},
 					},
 				},
@@ -2015,7 +2023,7 @@ func Test_getTags(t *testing.T) {
 					User: types.User{
 						Name: "joe",
 					},
-					HostInfo: types.HostInfo{
+					Hostinfo: &tailcfg.Hostinfo{
 						RequestTags: []string{"tag:invalid", "very-invalid"},
 					},
 				},
@@ -2056,10 +2064,6 @@ func Test_getTags(t *testing.T) {
 }
 
 func Test_getFilteredByACLPeers(t *testing.T) {
-	ipComparer := cmp.Comparer(func(x, y netip.Addr) bool {
-		return x.Compare(y) == 0
-	})
-
 	type args struct {
 		nodes types.Nodes
 		rules []tailcfg.FilterRule
@@ -2723,7 +2727,7 @@ func Test_getFilteredByACLPeers(t *testing.T) {
 				tt.args.nodes,
 				tt.args.rules,
 			)
-			if diff := cmp.Diff(tt.want, got, ipComparer); diff != "" {
+			if diff := cmp.Diff(tt.want, got, util.Comparers...); diff != "" {
 				t.Errorf("FilterNodesByACL() unexpected result (-want +got):\n%s", diff)
 			}
 		})
@@ -2986,9 +2990,6 @@ func TestValidExpandTagOwnersInSources(t *testing.T) {
 
 	node := &types.Node{
 		ID:          0,
-		MachineKey:  "foo",
-		NodeKey:     "bar",
-		DiscoKey:    "faa",
 		Hostname:    "testnodes",
 		IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.1")},
 		UserID:      0,
@@ -2996,7 +2997,7 @@ func TestValidExpandTagOwnersInSources(t *testing.T) {
 			Name: "user1",
 		},
 		RegisterMethod: util.RegisterMethodAuthKey,
-		HostInfo:       types.HostInfo(hostInfo),
+		Hostinfo:       &hostInfo,
 	}
 
 	pol := &ACLPolicy{
@@ -3041,9 +3042,6 @@ func TestInvalidTagValidUser(t *testing.T) {
 
 	node := &types.Node{
 		ID:          1,
-		MachineKey:  "12345",
-		NodeKey:     "bar",
-		DiscoKey:    "faa",
 		Hostname:    "testnodes",
 		IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.1")},
 		UserID:      1,
@@ -3051,7 +3049,7 @@ func TestInvalidTagValidUser(t *testing.T) {
 			Name: "user1",
 		},
 		RegisterMethod: util.RegisterMethodAuthKey,
-		HostInfo:       types.HostInfo(hostInfo),
+		Hostinfo:       &hostInfo,
 	}
 
 	pol := &ACLPolicy{
@@ -3095,9 +3093,6 @@ func TestValidExpandTagOwnersInDestinations(t *testing.T) {
 
 	node := &types.Node{
 		ID:          1,
-		MachineKey:  "12345",
-		NodeKey:     "bar",
-		DiscoKey:    "faa",
 		Hostname:    "testnodes",
 		IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.1")},
 		UserID:      1,
@@ -3105,7 +3100,7 @@ func TestValidExpandTagOwnersInDestinations(t *testing.T) {
 			Name: "user1",
 		},
 		RegisterMethod: util.RegisterMethodAuthKey,
-		HostInfo:       types.HostInfo(hostInfo),
+		Hostinfo:       &hostInfo,
 	}
 
 	pol := &ACLPolicy{
@@ -3159,9 +3154,6 @@ func TestValidTagInvalidUser(t *testing.T) {
 
 	node := &types.Node{
 		ID:          1,
-		MachineKey:  "12345",
-		NodeKey:     "bar",
-		DiscoKey:    "faa",
 		Hostname:    "webserver",
 		IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.1")},
 		UserID:      1,
@@ -3169,7 +3161,7 @@ func TestValidTagInvalidUser(t *testing.T) {
 			Name: "user1",
 		},
 		RegisterMethod: util.RegisterMethodAuthKey,
-		HostInfo:       types.HostInfo(hostInfo),
+		Hostinfo:       &hostInfo,
 	}
 
 	hostInfo2 := tailcfg.Hostinfo{
@@ -3179,9 +3171,6 @@ func TestValidTagInvalidUser(t *testing.T) {
 
 	nodes2 := &types.Node{
 		ID:          2,
-		MachineKey:  "56789",
-		NodeKey:     "bar2",
-		DiscoKey:    "faab",
 		Hostname:    "user",
 		IPAddresses: types.NodeAddresses{netip.MustParseAddr("100.64.0.2")},
 		UserID:      1,
@@ -3189,7 +3178,7 @@ func TestValidTagInvalidUser(t *testing.T) {
 			Name: "user1",
 		},
 		RegisterMethod: util.RegisterMethodAuthKey,
-		HostInfo:       types.HostInfo(hostInfo2),
+		Hostinfo:       &hostInfo2,
 	}
 
 	pol := &ACLPolicy{

hscontrol/poll.go

@@ -8,8 +8,8 @@ import (
 
 	"github.com/juanfont/headscale/hscontrol/mapper"
 	"github.com/juanfont/headscale/hscontrol/types"
-	"github.com/juanfont/headscale/hscontrol/util"
 	"github.com/rs/zerolog/log"
+	xslices "golang.org/x/exp/slices"
 	"tailscale.com/tailcfg"
 )
 
@@ -26,35 +26,32 @@ type UpdateNode func()
 func logPollFunc(
 	mapRequest tailcfg.MapRequest,
 	node *types.Node,
-	isNoise bool,
 ) (func(string), func(error, string)) {
 	return func(msg string) {
 			log.Info().
 				Caller().
-				Bool("noise", isNoise).
 				Bool("readOnly", mapRequest.ReadOnly).
 				Bool("omitPeers", mapRequest.OmitPeers).
 				Bool("stream", mapRequest.Stream).
-				Str("node_key", node.NodeKey).
+				Str("node_key", node.NodeKey.ShortString()).
 				Str("node", node.Hostname).
 				Msg(msg)
 		},
 		func(err error, msg string) {
 			log.Error().
 				Caller().
-				Bool("noise", isNoise).
 				Bool("readOnly", mapRequest.ReadOnly).
 				Bool("omitPeers", mapRequest.OmitPeers).
 				Bool("stream", mapRequest.Stream).
-				Str("node_key", node.NodeKey).
+				Str("node_key", node.NodeKey.ShortString()).
 				Str("node", node.Hostname).
 				Err(err).
 				Msg(msg)
 		}
 }
 
-// handlePoll is the common code for the legacy and Noise protocols to
+// handlePoll ensures the node gets the appropriate updates from either
-// managed the poll loop.
+// polling or immediate responses.
 //
 //nolint:gocyclo
 func (h *Headscale) handlePoll(
@@ -62,12 +59,10 @@ func (h *Headscale) handlePoll(
 	ctx context.Context,
 	node *types.Node,
 	mapRequest tailcfg.MapRequest,
-	isNoise bool,
-	capVer tailcfg.CapabilityVersion,
 ) {
-	logInfo, logErr := logPollFunc(mapRequest, node, isNoise)
+	logInfo, logErr := logPollFunc(mapRequest, node)
 
-	// This is the mechanism where the node gives us inforamtion about its
+	// This is the mechanism where the node gives us information about its
 	// current configuration.
 	//
 	// If OmitPeers is true, Stream is false, and ReadOnly is false,
@@ -75,24 +70,95 @@ func (h *Headscale) handlePoll(
 	// breaking existing long-polling (Stream == true) connections.
 	// In this case, the server can omit the entire response; the client
 	// only checks the HTTP response status code.
+	// TODO(kradalby): remove ReadOnly when we only support capVer 68+
 	if mapRequest.OmitPeers && !mapRequest.Stream && !mapRequest.ReadOnly {
 		log.Info().
 			Caller().
-			Bool("noise", isNoise).
 			Bool("readOnly", mapRequest.ReadOnly).
 			Bool("omitPeers", mapRequest.OmitPeers).
 			Bool("stream", mapRequest.Stream).
-			Str("node_key", node.NodeKey).
+			Str("node_key", node.NodeKey.ShortString()).
 			Str("node", node.Hostname).
-			Strs("endpoints", node.Endpoints).
+			Int("cap_ver", int(mapRequest.Version)).
-			Msg("Received endpoint update")
+			Msg("Received update")
 
-		now := time.Now().UTC()
+		change := node.PeerChangeFromMapRequest(mapRequest)
-		node.LastSeen = &now
+
-		node.Hostname = mapRequest.Hostinfo.Hostname
+		online := h.nodeNotifier.IsConnected(node.MachineKey)
-		node.HostInfo = types.HostInfo(*mapRequest.Hostinfo)
+		change.Online = &online
-		node.DiscoKey = util.DiscoPublicKeyStripPrefix(mapRequest.DiscoKey)
+
-		node.Endpoints = mapRequest.Endpoints
+		node.ApplyPeerChange(&change)
+
+		hostInfoChange := node.Hostinfo.Equal(mapRequest.Hostinfo)
+
+		logTracePeerChange(node.Hostname, hostInfoChange, &change)
+
+		// Check if the Hostinfo of the node has changed.
+		// If it has changed, check if there has been a change tod
+		// the routable IPs of the host and update update them in
+		// the database. Then send a Changed update
+		// (containing the whole node object) to peers to inform about
+		// the route change.
+		// If the hostinfo has changed, but not the routes, just update
+		// hostinfo and let the function continue.
+		if !hostInfoChange {
+			oldRoutes := node.Hostinfo.RoutableIPs
+			newRoutes := mapRequest.Hostinfo.RoutableIPs
+
+			oldServicesCount := len(node.Hostinfo.Services)
+			newServicesCount := len(mapRequest.Hostinfo.Services)
+
+			node.Hostinfo = mapRequest.Hostinfo
+
+			sendUpdate := false
+
+			// Route changes come as part of Hostinfo, which means that
+			// when an update comes, the Node Route logic need to run.
+			// This will require a "change" in comparison to a "patch",
+			// which is more costly.
+			if !xslices.Equal(oldRoutes, newRoutes) {
+				var err error
+				sendUpdate, err = h.db.SaveNodeRoutes(node)
+				if err != nil {
+					logErr(err, "Error processing node routes")
+					http.Error(writer, "", http.StatusInternalServerError)
+
+					return
+				}
+			}
+
+			// Services is mostly useful for discovery and not critical,
+			// except for peerapi, which is how nodes talk to eachother.
+			// If peerapi was not part of the initial mapresponse, we
+			// need to make sure its sent out later as it is needed for
+			// Taildrop.
+			// TODO(kradalby): Length comparison is a bit naive, replace.
+			if oldServicesCount != newServicesCount {
+				sendUpdate = true
+			}
+
+			if sendUpdate {
+				if err := h.db.NodeSave(node); err != nil {
+					logErr(err, "Failed to persist/update node in the database")
+					http.Error(writer, "", http.StatusInternalServerError)
+
+					return
+				}
+
+				stateUpdate := types.StateUpdate{
+					Type:        types.StatePeerChanged,
+					ChangeNodes: types.Nodes{node},
+					Message:     "called from handlePoll -> update -> new hostinfo",
+				}
+				if stateUpdate.Valid() {
+					h.nodeNotifier.NotifyWithIgnore(
+						stateUpdate,
+						node.MachineKey.String())
+				}
+
+				return
+			}
+		}
 
 		if err := h.db.NodeSave(node); err != nil {
 			logErr(err, "Failed to persist/update node in the database")
@@ -101,20 +167,15 @@ func (h *Headscale) handlePoll(
 			return
 		}
 
-		err := h.db.SaveNodeRoutes(node)
+		stateUpdate := types.StateUpdate{
-		if err != nil {
+			Type:          types.StatePeerChangedPatch,
-			logErr(err, "Error processing node routes")
+			ChangePatches: []*tailcfg.PeerChange{&change},
-			http.Error(writer, "", http.StatusInternalServerError)
+		}
-
+		if stateUpdate.Valid() {
-			return
+			h.nodeNotifier.NotifyWithIgnore(
+				stateUpdate,
+				node.MachineKey.String())
 		}
-
-		h.nodeNotifier.NotifyWithIgnore(
-			types.StateUpdate{
-				Type:    types.StatePeerChanged,
-				Changed: types.Nodes{node},
-			},
-			node.MachineKey)
 
 		writer.WriteHeader(http.StatusOK)
 		if f, ok := writer.(http.Flusher); ok {
@@ -122,7 +183,7 @@ func (h *Headscale) handlePoll(
 		}
 
 		return
-
+	} else if mapRequest.OmitPeers && !mapRequest.Stream && mapRequest.ReadOnly {
 		// ReadOnly is whether the client just wants to fetch the
 		// MapResponse, without updating their Endpoints. The
 		// Endpoints field will be ignored and LastSeen will not be
@@ -131,7 +192,7 @@ func (h *Headscale) handlePoll(
 		// The intended use is for clients to discover the DERP map at
 		// start-up before their first real endpoint update.
 	} else if mapRequest.OmitPeers && !mapRequest.Stream && mapRequest.ReadOnly {
-		h.handleLiteRequest(writer, node, mapRequest, isNoise, capVer)
+		h.handleLiteRequest(writer, node, mapRequest)
 
 		return
 	} else if mapRequest.OmitPeers && mapRequest.Stream {
@@ -140,12 +201,39 @@ func (h *Headscale) handlePoll(
 		return
 	}
 
-	now := time.Now().UTC()
+	change := node.PeerChangeFromMapRequest(mapRequest)
-	node.LastSeen = &now
+
-	node.Hostname = mapRequest.Hostinfo.Hostname
+	// A stream is being set up, the node is Online
-	node.HostInfo = types.HostInfo(*mapRequest.Hostinfo)
+	online := true
-	node.DiscoKey = util.DiscoPublicKeyStripPrefix(mapRequest.DiscoKey)
+	change.Online = &online
-	node.Endpoints = mapRequest.Endpoints
+
+	node.ApplyPeerChange(&change)
+
+	// Only save HostInfo if changed, update routes if changed
+	// TODO(kradalby): Remove when capver is over 68
+	if !node.Hostinfo.Equal(mapRequest.Hostinfo) {
+		oldRoutes := node.Hostinfo.RoutableIPs
+		newRoutes := mapRequest.Hostinfo.RoutableIPs
+
+		node.Hostinfo = mapRequest.Hostinfo
+
+		if !xslices.Equal(oldRoutes, newRoutes) {
+			_, err := h.db.SaveNodeRoutes(node)
+			if err != nil {
+				logErr(err, "Error processing node routes")
+				http.Error(writer, "", http.StatusInternalServerError)
+
+				return
+			}
+		}
+	}
+
+	if err := h.db.NodeSave(node); err != nil {
+		logErr(err, "Failed to persist/update node in the database")
+		http.Error(writer, "", http.StatusInternalServerError)
+
+		return
+	}
 
 	// When a node connects to control, list the peers it has at
 	// that given point, further updates are kept in memory in
@@ -159,12 +247,14 @@ func (h *Headscale) handlePoll(
 		return
 	}
 
+	for _, peer := range peers {
+		online := h.nodeNotifier.IsConnected(peer.MachineKey)
+		peer.IsOnline = &online
+	}
+
 	mapp := mapper.NewMapper(
 		node,
 		peers,
-		h.privateKey2019,
-		isNoise,
-		capVer,
 		h.DERPMap,
 		h.cfg.BaseDomain,
 		h.cfg.DNSConfig,
@@ -172,11 +262,6 @@ func (h *Headscale) handlePoll(
 		h.cfg.RandomizeClientPort,
 	)
 
-	err = h.db.SaveNodeRoutes(node)
-	if err != nil {
-		logErr(err, "Error processing node routes")
-	}
-
 	// update ACLRules with peer informations (to update server tags if necessary)
 	if h.ACLPolicy != nil {
 		// update routes with peer information
@@ -186,14 +271,6 @@ func (h *Headscale) handlePoll(
 		}
 	}
 
-	// TODO(kradalby): Save specific stuff, not whole object.
-	if err := h.db.NodeSave(node); err != nil {
-		logErr(err, "Failed to persist/update node in the database")
-		http.Error(writer, "", http.StatusInternalServerError)
-
-		return
-	}
-
 	logInfo("Sending initial map")
 
 	mapResp, err := mapp.FullMapResponse(mapRequest, node, h.ACLPolicy)
@@ -218,18 +295,26 @@ func (h *Headscale) handlePoll(
 		return
 	}
 
-	h.nodeNotifier.NotifyWithIgnore(
+	stateUpdate := types.StateUpdate{
-		types.StateUpdate{
+		Type:        types.StatePeerChanged,
-			Type:    types.StatePeerChanged,
+		ChangeNodes: types.Nodes{node},
-			Changed: types.Nodes{node},
+		Message:     "called from handlePoll -> new node added",
-		},
+	}
-		node.MachineKey)
+	if stateUpdate.Valid() {
+		h.nodeNotifier.NotifyWithIgnore(
+			stateUpdate,
+			node.MachineKey.String())
+	}
 
 	// Set up the client stream
 	h.pollNetMapStreamWG.Add(1)
 	defer h.pollNetMapStreamWG.Done()
 
-	updateChan := make(chan types.StateUpdate)
+	// Use a buffered channel in case a node is not fully ready
+	// to receive a message to make sure we dont block the entire
+	// notifier.
+	// 12 is arbitrarily chosen.
+	updateChan := make(chan types.StateUpdate, 12)
 	defer closeChanWithLog(updateChan, node.Hostname, "updateChan")
 
 	// Register the node's update channel
@@ -243,6 +328,10 @@ func (h *Headscale) handlePoll(
 	ctx, cancel := context.WithCancel(ctx)
 	defer cancel()
 
+	if len(node.Routes) > 0 {
+		go h.db.EnsureFailoverRouteIsAvailable(node)
+	}
+
 	for {
 		logInfo("Waiting for update on stream channel")
 		select {
@@ -272,14 +361,7 @@ func (h *Headscale) handlePoll(
 			// One alternative is to split these different channels into
 			// goroutines, but then you might have a problem without a lock
 			// if a keepalive is written at the same time as an update.
-			go func() {
+			go h.updateNodeOnlineStatus(true, node)
-				err = h.db.UpdateLastSeen(node)
-				if err != nil {
-					logErr(err, "Cannot update node LastSeen")
-
-					return
-				}
-			}()
 
 		case update := <-updateChan:
 			logInfo("Received update")
@@ -289,18 +371,43 @@ func (h *Headscale) handlePoll(
 			var err error
 
 			switch update.Type {
+			case types.StateFullUpdate:
+				logInfo("Sending Full MapResponse")
+
+				data, err = mapp.FullMapResponse(mapRequest, node, h.ACLPolicy)
 			case types.StatePeerChanged:
-				logInfo("Sending PeerChanged MapResponse")
+				logInfo(fmt.Sprintf("Sending Changed MapResponse: %s", update.Message))
-				data, err = mapp.PeerChangedResponse(mapRequest, node, update.Changed, h.ACLPolicy)
+
+				for _, node := range update.ChangeNodes {
+					// If a node is not reported to be online, it might be
+					// because the value is outdated, check with the notifier.
+					// However, if it is set to Online, and not in the notifier,
+					// this might be because it has announced itself, but not
+					// reached the stage to actually create the notifier channel.
+					if node.IsOnline != nil && !*node.IsOnline {
+						isOnline := h.nodeNotifier.IsConnected(node.MachineKey)
+						node.IsOnline = &isOnline
+					}
+				}
+
+				data, err = mapp.PeerChangedResponse(mapRequest, node, update.ChangeNodes, h.ACLPolicy, update.Message)
+			case types.StatePeerChangedPatch:
+				logInfo("Sending PeerChangedPatch MapResponse")
+				data, err = mapp.PeerChangedPatchResponse(mapRequest, node, update.ChangePatches, h.ACLPolicy)
 			case types.StatePeerRemoved:
 				logInfo("Sending PeerRemoved MapResponse")
 				data, err = mapp.PeerRemovedResponse(mapRequest, node, update.Removed)
+			case types.StateSelfUpdate:
+				if len(update.ChangeNodes) == 1 {
+					logInfo("Sending SelfUpdate MapResponse")
+					node = update.ChangeNodes[0]
+					data, err = mapp.LiteMapResponse(mapRequest, node, h.ACLPolicy)
+				} else {
+					logInfo("SelfUpdate contained too many nodes, this is likely a bug in the code, please report.")
+				}
 			case types.StateDERPUpdated:
 				logInfo("Sending DERPUpdate MapResponse")
 				data, err = mapp.DERPMapResponse(mapRequest, node, update.DERPMap)
-			case types.StateFullUpdate:
-				logInfo("Sending Full MapResponse")
-				data, err = mapp.FullMapResponse(mapRequest, node, h.ACLPolicy)
 			}
 
 			if err != nil {
@@ -309,55 +416,45 @@ func (h *Headscale) handlePoll(
 				return
 			}
 
-			_, err = writer.Write(data)
+			// Only send update if there is change
-			if err != nil {
+			if data != nil {
-				logErr(err, "Could not write the map response")
+				_, err = writer.Write(data)
-
-				updateRequestsSentToNode.WithLabelValues(node.User.Name, node.Hostname, "failed").
-					Inc()
-
-				return
-			}
-
-			if flusher, ok := writer.(http.Flusher); ok {
-				flusher.Flush()
-			} else {
-				log.Error().Msg("Failed to create http flusher")
-
-				return
-			}
-
-			// See comment in keepAliveTicker
-			go func() {
-				err = h.db.UpdateLastSeen(node)
 				if err != nil {
-					logErr(err, "Cannot update node LastSeen")
+					logErr(err, "Could not write the map response")
+
+					updateRequestsSentToNode.WithLabelValues(node.User.Name, node.Hostname, "failed").
+						Inc()
 
 					return
 				}
-			}()
 
-			log.Info().
+				if flusher, ok := writer.(http.Flusher); ok {
-				Caller().
+					flusher.Flush()
-				Bool("noise", isNoise).
+				} else {
-				Bool("readOnly", mapRequest.ReadOnly).
+					log.Error().Msg("Failed to create http flusher")
-				Bool("omitPeers", mapRequest.OmitPeers).
+
-				Bool("stream", mapRequest.Stream).
+					return
-				Str("node_key", node.NodeKey).
+				}
-				Str("node", node.Hostname).
+
-				TimeDiff("timeSpent", time.Now(), now).
+				log.Info().
-				Msg("update sent")
+					Caller().
+					Bool("readOnly", mapRequest.ReadOnly).
+					Bool("omitPeers", mapRequest.OmitPeers).
+					Bool("stream", mapRequest.Stream).
+					Str("node_key", node.NodeKey.ShortString()).
+					Str("machine_key", node.MachineKey.ShortString()).
+					Str("node", node.Hostname).
+					TimeDiff("timeSpent", time.Now(), now).
+					Msg("update sent")
+			}
+
 		case <-ctx.Done():
 			logInfo("The client has closed the connection")
 
-			go func() {
+			go h.updateNodeOnlineStatus(false, node)
-				err = h.db.UpdateLastSeen(node)
-				if err != nil {
-					logErr(err, "Cannot update node LastSeen")
 
-					return
+			// Failover the node's routes if any.
-				}
+			go h.db.FailoverNodeRoutesWithNotify(node)
-			}()
 
 			// The connection has been closed, so we can stop polling.
 			return
@@ -370,6 +467,36 @@ func (h *Headscale) handlePoll(
 	}
 }
 
+// updateNodeOnlineStatus records the last seen status of a node and notifies peers
+// about change in their online/offline status.
+// It takes a StateUpdateType of either StatePeerOnlineChanged or StatePeerOfflineChanged.
+func (h *Headscale) updateNodeOnlineStatus(online bool, node *types.Node) {
+	now := time.Now()
+
+	node.LastSeen = &now
+
+	statusUpdate := types.StateUpdate{
+		Type: types.StatePeerChangedPatch,
+		ChangePatches: []*tailcfg.PeerChange{
+			{
+				NodeID:   tailcfg.NodeID(node.ID),
+				Online:   &online,
+				LastSeen: &now,
+			},
+		},
+	}
+	if statusUpdate.Valid() {
+		h.nodeNotifier.NotifyWithIgnore(statusUpdate, node.MachineKey.String())
+	}
+
+	err := h.db.UpdateLastSeen(node)
+	if err != nil {
+		log.Error().Err(err).Msg("Cannot update node LastSeen")
+
+		return
+	}
+}
+
 func closeChanWithLog[C chan []byte | chan struct{} | chan types.StateUpdate](channel C, node, name string) {
 	log.Trace().
 		Str("handler", "PollNetMap").
@@ -384,19 +511,12 @@ func (h *Headscale) handleLiteRequest(
 	writer http.ResponseWriter,
 	node *types.Node,
 	mapRequest tailcfg.MapRequest,
-	isNoise bool,
-	capVer tailcfg.CapabilityVersion,
 ) {
-	logInfo, logErr := logPollFunc(mapRequest, node, isNoise)
+	logInfo, logErr := logPollFunc(mapRequest, node)
 
 	mapp := mapper.NewMapper(
 		node,
-		// TODO(kradalby): It might not be acceptable to send
-		// an empty peer list here.
 		types.Nodes{},
-		h.privateKey2019,
-		isNoise,
-		capVer,
 		h.DERPMap,
 		h.cfg.BaseDomain,
 		h.cfg.DNSConfig,
@@ -421,3 +541,38 @@ func (h *Headscale) handleLiteRequest(
 		logErr(err, "Failed to write response")
 	}
 }
+
+func logTracePeerChange(hostname string, hostinfoChange bool, change *tailcfg.PeerChange) {
+	trace := log.Trace().Str("node_id", change.NodeID.String()).Str("hostname", hostname)
+
+	if change.Key != nil {
+		trace = trace.Str("node_key", change.Key.ShortString())
+	}
+
+	if change.DiscoKey != nil {
+		trace = trace.Str("disco_key", change.DiscoKey.ShortString())
+	}
+
+	if change.Online != nil {
+		trace = trace.Bool("online", *change.Online)
+	}
+
+	if change.Endpoints != nil {
+		eps := make([]string, len(change.Endpoints))
+		for idx, ep := range change.Endpoints {
+			eps[idx] = ep.String()
+		}
+
+		trace = trace.Strs("endpoints", eps)
+	}
+
+	if hostinfoChange {
+		trace = trace.Bool("hostinfo_changed", hostinfoChange)
+	}
+
+	if change.DERPRegion != 0 {
+		trace = trace.Int("derp_region", change.DERPRegion)
+	}
+
+	trace.Time("last_seen", *change.LastSeen).Msg("PeerChange received")
+}

hscontrol/poll_legacy.go

@@ -1,108 +0,0 @@
-//go:build ts2019
-
-package hscontrol
-
-import (
-	"errors"
-	"io"
-	"net/http"
-
-	"github.com/gorilla/mux"
-	"github.com/juanfont/headscale/hscontrol/util"
-	"github.com/rs/zerolog/log"
-	"gorm.io/gorm"
-	"tailscale.com/tailcfg"
-	"tailscale.com/types/key"
-)
-
-// PollNetMapHandler takes care of /machine/:id/map
-//
-// This is the busiest endpoint, as it keeps the HTTP long poll that updates
-// the clients when something in the network changes.
-//
-// The clients POST stuff like HostInfo and their Endpoints here, but
-// only after their first request (marked with the ReadOnly field).
-//
-// At this moment the updates are sent in a quite horrendous way, but they kinda work.
-func (h *Headscale) PollNetMapHandler(
-	writer http.ResponseWriter,
-	req *http.Request,
-) {
-	vars := mux.Vars(req)
-	machineKeyStr, ok := vars["mkey"]
-	if !ok || machineKeyStr == "" {
-		log.Error().
-			Str("handler", "PollNetMap").
-			Msg("No machine key in request")
-		http.Error(writer, "No machine key in request", http.StatusBadRequest)
-
-		return
-	}
-	log.Trace().
-		Str("handler", "PollNetMap").
-		Str("id", machineKeyStr).
-		Msg("PollNetMapHandler called")
-	body, _ := io.ReadAll(req.Body)
-
-	var machineKey key.MachinePublic
-	err := machineKey.UnmarshalText([]byte(util.MachinePublicKeyEnsurePrefix(machineKeyStr)))
-	if err != nil {
-		log.Error().
-			Str("handler", "PollNetMap").
-			Err(err).
-			Msg("Cannot parse client key")
-
-		http.Error(writer, "Cannot parse client key", http.StatusBadRequest)
-
-		return
-	}
-	mapRequest := tailcfg.MapRequest{}
-	err = util.DecodeAndUnmarshalNaCl(body, &mapRequest, &machineKey, h.privateKey2019)
-	if err != nil {
-		log.Error().
-			Str("handler", "PollNetMap").
-			Err(err).
-			Msg("Cannot decode message")
-		http.Error(writer, "Cannot decode message", http.StatusBadRequest)
-
-		return
-	}
-
-	node, err := h.db.GetNodeByMachineKey(machineKey)
-	if err != nil {
-		if errors.Is(err, gorm.ErrRecordNotFound) {
-			log.Warn().
-				Str("handler", "PollNetMap").
-				Msgf("Ignoring request, cannot find node with key %s", machineKey.String())
-
-			http.Error(writer, "", http.StatusUnauthorized)
-
-			return
-		}
-		log.Error().
-			Str("handler", "PollNetMap").
-			Msgf("Failed to fetch node from the database with Machine key: %s", machineKey.String())
-		http.Error(writer, "", http.StatusInternalServerError)
-
-		return
-	}
-
-	log.Trace().
-		Str("handler", "PollNetMap").
-		Str("id", machineKeyStr).
-		Str("node", node.Hostname).
-		Msg("A node is sending a MapRequest via legacy protocol")
-
-	capVer, err := parseCabailityVersion(req)
-	if err != nil && !errors.Is(err, ErrNoCapabilityVersion) {
-		log.Error().
-			Caller().
-			Err(err).
-			Msg("failed to parse capVer")
-		http.Error(writer, "Internal error", http.StatusInternalServerError)
-
-		return
-	}
-
-	h.handlePoll(writer, req.Context(), node, mapRequest, false, capVer)
-}

hscontrol/poll_noise.go

@@ -12,6 +12,10 @@ import (
 	"tailscale.com/types/key"
 )
 
+const (
+	MinimumCapVersion tailcfg.CapabilityVersion = 56
+)
+
 // NoisePollNetMapHandler takes care of /machine/:id/map using the Noise protocol
 //
 // This is the busiest endpoint, as it keeps the HTTP long poll that updates
@@ -47,6 +51,18 @@ func (ns *noiseServer) NoisePollNetMapHandler(
 		return
 	}
 
+	// Reject unsupported versions
+	if mapRequest.Version < MinimumCapVersion {
+		log.Info().
+			Caller().
+			Int("min_version", int(MinimumCapVersion)).
+			Int("client_version", int(mapRequest.Version)).
+			Msg("unsupported client connected")
+		http.Error(writer, "Internal error", http.StatusBadRequest)
+
+		return
+	}
+
 	ns.nodeKey = mapRequest.NodeKey
 
 	node, err := ns.headscale.db.GetNodeByAnyKey(
@@ -73,20 +89,8 @@ func (ns *noiseServer) NoisePollNetMapHandler(
 	log.Debug().
 		Str("handler", "NoisePollNetMap").
 		Str("node", node.Hostname).
+		Int("cap_ver", int(mapRequest.Version)).
 		Msg("A node sending a MapRequest with Noise protocol")
 
-	capVer, err := parseCabailityVersion(req)
+	ns.headscale.handlePoll(writer, req.Context(), node, mapRequest)
-	if err != nil && !errors.Is(err, ErrNoCapabilityVersion) {
-		log.Error().
-			Caller().
-			Err(err).
-			Msg("failed to parse capVer")
-		http.Error(writer, "Internal error", http.StatusInternalServerError)
-
-		return
-	}
-
-	// TODO(kradalby): since we are now passing capVer, we could arguably stop passing
-	// isNoise, and rather have a isNoise function that takes capVer
-	ns.headscale.handlePoll(writer, req.Context(), node, mapRequest, true, capVer)
 }

hscontrol/suite_test.go

@@ -40,7 +40,6 @@ func (s *Suite) ResetDB(c *check.C) {
 		c.Fatal(err)
 	}
 	cfg := types.Config{
-		PrivateKeyPath:      tmpDir + "/private.key",
 		NoisePrivateKeyPath: tmpDir + "/noise_private.key",
 		DBtype:              "sqlite3",
 		DBpath:              tmpDir + "/headscale_test.db",

hscontrol/tailsql.go

@@ -0,0 +1,99 @@
+package hscontrol
+
+import (
+	"context"
+	"fmt"
+	"net/http"
+	"os"
+
+	"github.com/tailscale/tailsql/server/tailsql"
+	"tailscale.com/tsnet"
+	"tailscale.com/tsweb"
+	"tailscale.com/types/logger"
+)
+
+func runTailSQLService(ctx context.Context, logf logger.Logf, stateDir, dbPath string) error {
+	opts := tailsql.Options{
+		Hostname: "tailsql-headscale",
+		StateDir: stateDir,
+		Sources: []tailsql.DBSpec{
+			{
+				Source: "headscale",
+				Label:  "headscale - sqlite",
+				Driver: "sqlite",
+				URL:    fmt.Sprintf("file:%s?mode=ro", dbPath),
+				Named: map[string]string{
+					"schema": `select * from sqlite_schema`,
+				},
+			},
+		},
+	}
+
+	tsNode := &tsnet.Server{
+		Dir:      os.ExpandEnv(opts.StateDir),
+		Hostname: opts.Hostname,
+		Logf:     logger.Discard,
+	}
+	// if *doDebugLog {
+	// 	tsNode.Logf = logf
+	// }
+	defer tsNode.Close()
+
+	logf("Starting tailscale (hostname=%q)", opts.Hostname)
+	lc, err := tsNode.LocalClient()
+	if err != nil {
+		return fmt.Errorf("connect local client: %w", err)
+	}
+	opts.LocalClient = lc // for authentication
+
+	// Make sure the Tailscale node starts up. It might not, if it is a new node
+	// and the user did not provide an auth key.
+	if st, err := tsNode.Up(ctx); err != nil {
+		return fmt.Errorf("starting tailscale: %w", err)
+	} else {
+		logf("tailscale started, node state %q", st.BackendState)
+	}
+
+	// Reaching here, we have a running Tailscale node, now we can set up the
+	// HTTP and/or HTTPS plumbing for TailSQL itself.
+	tsql, err := tailsql.NewServer(opts)
+	if err != nil {
+		return fmt.Errorf("creating tailsql server: %w", err)
+	}
+
+	lst, err := tsNode.Listen("tcp", ":80")
+	if err != nil {
+		return fmt.Errorf("listen port 80: %w", err)
+	}
+
+	if opts.ServeHTTPS {
+		// When serving TLS, add a redirect from HTTP on port 80 to HTTPS on 443.
+		certDomains := tsNode.CertDomains()
+		if len(certDomains) == 0 {
+			fmt.Errorf("no cert domains available for HTTPS")
+		}
+		base := "https://" + certDomains[0]
+		go http.Serve(lst, http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+			target := base + r.RequestURI
+			http.Redirect(w, r, target, http.StatusPermanentRedirect)
+		}))
+		// log.Printf("Redirecting HTTP to HTTPS at %q", base)
+
+		// For the real service, start a separate listener.
+		// Note: Replaces the port 80 listener.
+		var err error
+		lst, err = tsNode.ListenTLS("tcp", ":443")
+		if err != nil {
+			return fmt.Errorf("listen TLS: %w", err)
+		}
+		logf("enabled serving via HTTPS")
+	}
+
+	mux := tsql.NewMux()
+	tsweb.Debugger(mux)
+	go http.Serve(lst, mux)
+	logf("ailSQL started")
+	<-ctx.Done()
+	logf("TailSQL shutting down...")
+	return tsNode.Close()
+}

hscontrol/types/common.go

@@ -12,33 +12,6 @@ import (
 
 var ErrCannotParsePrefix = errors.New("cannot parse prefix")
 
-// This is a "wrapper" type around tailscales
-// Hostinfo to allow us to add database "serialization"
-// methods. This allows us to use a typed values throughout
-// the code and not have to marshal/unmarshal and error
-// check all over the code.
-type HostInfo tailcfg.Hostinfo
-
-func (hi *HostInfo) Scan(destination interface{}) error {
-	switch value := destination.(type) {
-	case []byte:
-		return json.Unmarshal(value, hi)
-
-	case string:
-		return json.Unmarshal([]byte(value), hi)
-
-	default:
-		return fmt.Errorf("%w: unexpected data type %T", ErrNodeAddressesInvalid, destination)
-	}
-}
-
-// Value return json value, implement driver.Valuer interface.
-func (hi HostInfo) Value() (driver.Value, error) {
-	bytes, err := json.Marshal(hi)
-
-	return string(bytes), err
-}
-
 type IPPrefix netip.Prefix
 
 func (i *IPPrefix) Scan(destination interface{}) error {
@@ -111,20 +84,37 @@ type StateUpdateType int
 
 const (
 	StateFullUpdate StateUpdateType = iota
+	// StatePeerChanged is used for updates that needs
+	// to be calculated with all peers and all policy rules.
+	// This would typically be things that include tags, routes
+	// and similar.
 	StatePeerChanged
+	StatePeerChangedPatch
 	StatePeerRemoved
+	// StateSelfUpdate is used to indicate that the node
+	// has changed in control, and the client needs to be
+	// informed.
+	// The updated node is inside the ChangeNodes field
+	// which should have a length of one.
+	StateSelfUpdate
 	StateDERPUpdated
 )
 
 // StateUpdate is an internal message containing information about
 // a state change that has happened to the network.
+// If type is StateFullUpdate, all fields are ignored.
 type StateUpdate struct {
 	// The type of update
 	Type StateUpdateType
 
-	// Changed must be set when Type is StatePeerChanged and
+	// ChangeNodes must be set when Type is StatePeerAdded
-	// contain the Node IDs of nodes that have changed.
+	// and StatePeerChanged and contains the full node
-	Changed Nodes
+	// object for added nodes.
+	ChangeNodes Nodes
+
+	// ChangePatches must be set when Type is StatePeerChangedPatch
+	// and contains a populated PeerChange object.
+	ChangePatches []*tailcfg.PeerChange
 
 	// Removed must be set when Type is StatePeerRemoved and
 	// contain a list of the nodes that has been removed from
@@ -133,5 +123,40 @@ type StateUpdate struct {
 
 	// DERPMap must be set when Type is StateDERPUpdated and
 	// contain the new DERP Map.
-	DERPMap tailcfg.DERPMap
+	DERPMap *tailcfg.DERPMap
+
+	// Additional message for tracking origin or what being
+	// updated, useful for ambiguous updates like StatePeerChanged.
+	Message string
+}
+
+// Valid reports if a StateUpdate is correctly filled and
+// panics if the mandatory fields for a type is not
+// filled.
+// Reports true if valid.
+func (su *StateUpdate) Valid() bool {
+	switch su.Type {
+	case StatePeerChanged:
+		if su.ChangeNodes == nil {
+			panic("Mandatory field ChangeNodes is not set on StatePeerChanged update")
+		}
+	case StatePeerChangedPatch:
+		if su.ChangePatches == nil {
+			panic("Mandatory field ChangePatches is not set on StatePeerChangedPatch update")
+		}
+	case StatePeerRemoved:
+		if su.Removed == nil {
+			panic("Mandatory field Removed is not set on StatePeerRemove update")
+		}
+	case StateSelfUpdate:
+		if su.ChangeNodes == nil || len(su.ChangeNodes) != 1 {
+			panic("Mandatory field ChangeNodes is not set for StateSelfUpdate or has more than one node")
+		}
+	case StateDERPUpdated:
+		if su.DERPMap == nil {
+			panic("Mandatory field DERPMap is not set on StateDERPUpdated update")
+		}
+	}
+
+	return true
 }

hscontrol/types/config.go

@@ -41,7 +41,6 @@ type Config struct {
 	EphemeralNodeInactivityTimeout time.Duration
 	NodeUpdateCheckInterval        time.Duration
 	IPPrefixes                     []netip.Prefix
-	PrivateKeyPath                 string
 	NoisePrivateKeyPath            string
 	BaseDomain                     string
 	Log                            LogConfig
@@ -112,15 +111,16 @@ type OIDCConfig struct {
 }
 
 type DERPConfig struct {
-	ServerEnabled    bool
+	ServerEnabled        bool
-	ServerRegionID   int
+	ServerRegionID       int
-	ServerRegionCode string
+	ServerRegionCode     string
-	ServerRegionName string
+	ServerRegionName     string
-	STUNAddr         string
+	ServerPrivateKeyPath string
-	URLs             []url.URL
+	STUNAddr             string
-	Paths            []string
+	URLs                 []url.URL
-	AutoUpdate       bool
+	Paths                []string
-	UpdateFrequency  time.Duration
+	AutoUpdate           bool
+	UpdateFrequency      time.Duration
 }
 
 type LogTailConfig struct {
@@ -294,6 +294,7 @@ func GetDERPConfig() DERPConfig {
 	serverRegionCode := viper.GetString("derp.server.region_code")
 	serverRegionName := viper.GetString("derp.server.region_name")
 	stunAddr := viper.GetString("derp.server.stun_listen_addr")
+	privateKeyPath := util.AbsolutePathFromConfigPath(viper.GetString("derp.server.private_key_path"))
 
 	if serverEnabled && stunAddr == "" {
 		log.Fatal().
@@ -321,15 +322,16 @@ func GetDERPConfig() DERPConfig {
 	updateFrequency := viper.GetDuration("derp.update_frequency")
 
 	return DERPConfig{
-		ServerEnabled:    serverEnabled,
+		ServerEnabled:        serverEnabled,
-		ServerRegionID:   serverRegionID,
+		ServerRegionID:       serverRegionID,
-		ServerRegionCode: serverRegionCode,
+		ServerRegionCode:     serverRegionCode,
-		ServerRegionName: serverRegionName,
+		ServerRegionName:     serverRegionName,
-		STUNAddr:         stunAddr,
+		ServerPrivateKeyPath: privateKeyPath,
-		URLs:             urls,
+		STUNAddr:             stunAddr,
-		Paths:            paths,
+		URLs:                 urls,
-		AutoUpdate:       autoUpdate,
+		Paths:                paths,
-		UpdateFrequency:  updateFrequency,
+		AutoUpdate:           autoUpdate,
+		UpdateFrequency:      updateFrequency,
 	}
 }
 
@@ -590,9 +592,6 @@ func GetHeadscaleConfig() (*Config, error) {
 		DisableUpdateCheck: viper.GetBool("disable_check_updates"),
 
 		IPPrefixes: prefixes,
-		PrivateKeyPath: util.AbsolutePathFromConfigPath(
-			viper.GetString("private_key_path"),
-		),
 		NoisePrivateKeyPath: util.AbsolutePathFromConfigPath(
 			viper.GetString("noise.private_key_path"),
 		),

hscontrol/types/node.go

@@ -2,6 +2,7 @@ package types
 
 import (
 	"database/sql/driver"
+	"encoding/json"
 	"errors"
 	"fmt"
 	"net/netip"
@@ -11,24 +12,60 @@ import (
 
 	v1 "github.com/juanfont/headscale/gen/go/headscale/v1"
 	"github.com/juanfont/headscale/hscontrol/policy/matcher"
-	"github.com/juanfont/headscale/hscontrol/util"
+	"github.com/rs/zerolog/log"
 	"go4.org/netipx"
 	"google.golang.org/protobuf/types/known/timestamppb"
+	"gorm.io/gorm"
 	"tailscale.com/tailcfg"
 	"tailscale.com/types/key"
 )
 
 var (
 	ErrNodeAddressesInvalid = errors.New("failed to parse node addresses")
-	ErrHostnameTooLong      = errors.New("hostname too long")
+	ErrHostnameTooLong      = errors.New("hostname too long, cannot except 255 ASCII chars")
+	ErrNodeHasNoGivenName   = errors.New("node has no given name")
+	ErrNodeUserHasNoName    = errors.New("node user has no name")
 )
 
 // Node is a Headscale client.
 type Node struct {
-	ID          uint64 `gorm:"primary_key"`
+	ID uint64 `gorm:"primary_key"`
-	MachineKey  string `gorm:"type:varchar(64);unique_index"`
+
-	NodeKey     string
+	// MachineKeyDatabaseField is the string representation of MachineKey
-	DiscoKey    string
+	// it is _only_ used for reading and writing the key to the
+	// database and should not be used.
+	// Use MachineKey instead.
+	MachineKeyDatabaseField string            `gorm:"column:machine_key;unique_index"`
+	MachineKey              key.MachinePublic `gorm:"-"`
+
+	// NodeKeyDatabaseField is the string representation of NodeKey
+	// it is _only_ used for reading and writing the key to the
+	// database and should not be used.
+	// Use NodeKey instead.
+	NodeKeyDatabaseField string         `gorm:"column:node_key"`
+	NodeKey              key.NodePublic `gorm:"-"`
+
+	// DiscoKeyDatabaseField is the string representation of DiscoKey
+	// it is _only_ used for reading and writing the key to the
+	// database and should not be used.
+	// Use DiscoKey instead.
+	DiscoKeyDatabaseField string          `gorm:"column:disco_key"`
+	DiscoKey              key.DiscoPublic `gorm:"-"`
+
+	// EndpointsDatabaseField is the string list representation of Endpoints
+	// it is _only_ used for reading and writing the key to the
+	// database and should not be used.
+	// Use Endpoints instead.
+	EndpointsDatabaseField StringList       `gorm:"column:endpoints"`
+	Endpoints              []netip.AddrPort `gorm:"-"`
+
+	// EndpointsDatabaseField is the string list representation of Endpoints
+	// it is _only_ used for reading and writing the key to the
+	// database and should not be used.
+	// Use Endpoints instead.
+	HostinfoDatabaseField string            `gorm:"column:host_info"`
+	Hostinfo              *tailcfg.Hostinfo `gorm:"-"`
+
 	IPAddresses NodeAddresses
 
 	// Hostname represents the name given by the Tailscale
@@ -56,30 +93,19 @@ type Node struct {
 	LastSeen *time.Time
 	Expiry   *time.Time
 
-	HostInfo  HostInfo
-	Endpoints StringList
-
 	Routes []Route
 
 	CreatedAt time.Time
 	UpdatedAt time.Time
 	DeletedAt *time.Time
+
+	IsOnline *bool `gorm:"-"`
 }
 
 type (
 	Nodes []*Node
 )
 
-func (nodes Nodes) OnlineNodeMap() map[tailcfg.NodeID]bool {
-	ret := make(map[tailcfg.NodeID]bool)
-
-	for _, node := range nodes {
-		ret[tailcfg.NodeID(node.ID)] = node.IsOnline()
-	}
-
-	return ret
-}
-
 type NodeAddresses []netip.Addr
 
 func (na NodeAddresses) Sort() {
@@ -175,21 +201,6 @@ func (node Node) IsExpired() bool {
 	return time.Now().UTC().After(*node.Expiry)
 }
 
-// IsOnline returns if the node is connected to Headscale.
-// This is really a naive implementation, as we don't really see
-// if there is a working connection between the client and the server.
-func (node *Node) IsOnline() bool {
-	if node.LastSeen == nil {
-		return false
-	}
-
-	if node.IsExpired() {
-		return false
-	}
-
-	return node.LastSeen.After(time.Now().Add(-KeepAliveInterval))
-}
-
 // IsEphemeral returns if the node is registered as an Ephemeral node.
 // https://tailscale.com/kb/1111/ephemeral-nodes/
 func (node *Node) IsEphemeral() bool {
@@ -227,19 +238,89 @@ func (nodes Nodes) FilterByIP(ip netip.Addr) Nodes {
 	return found
 }
 
+// BeforeSave is a hook that ensures that some values that
+// cannot be directly marshalled into database values are stored
+// correctly in the database.
+// This currently means storing the keys as strings.
+func (node *Node) BeforeSave(tx *gorm.DB) error {
+	node.MachineKeyDatabaseField = node.MachineKey.String()
+	node.NodeKeyDatabaseField = node.NodeKey.String()
+	node.DiscoKeyDatabaseField = node.DiscoKey.String()
+
+	var endpoints StringList
+	for _, addrPort := range node.Endpoints {
+		endpoints = append(endpoints, addrPort.String())
+	}
+
+	node.EndpointsDatabaseField = endpoints
+
+	hi, err := json.Marshal(node.Hostinfo)
+	if err != nil {
+		return fmt.Errorf("failed to marshal Hostinfo to store in db: %w", err)
+	}
+	node.HostinfoDatabaseField = string(hi)
+
+	return nil
+}
+
+// AfterFind is a hook that ensures that Node objects fields that
+// has a different type in the database is unwrapped and populated
+// correctly.
+// This currently unmarshals all the keys, stored as strings, into
+// the proper types.
+func (node *Node) AfterFind(tx *gorm.DB) error {
+	var machineKey key.MachinePublic
+	if err := machineKey.UnmarshalText([]byte(node.MachineKeyDatabaseField)); err != nil {
+		return fmt.Errorf("failed to unmarshal machine key from db: %w", err)
+	}
+	node.MachineKey = machineKey
+
+	var nodeKey key.NodePublic
+	if err := nodeKey.UnmarshalText([]byte(node.NodeKeyDatabaseField)); err != nil {
+		return fmt.Errorf("failed to unmarshal node key from db: %w", err)
+	}
+	node.NodeKey = nodeKey
+
+	var discoKey key.DiscoPublic
+	if err := discoKey.UnmarshalText([]byte(node.DiscoKeyDatabaseField)); err != nil {
+		return fmt.Errorf("failed to unmarshal disco key from db: %w", err)
+	}
+	node.DiscoKey = discoKey
+
+	endpoints := make([]netip.AddrPort, len(node.EndpointsDatabaseField))
+	for idx, ep := range node.EndpointsDatabaseField {
+		addrPort, err := netip.ParseAddrPort(ep)
+		if err != nil {
+			return fmt.Errorf("failed to parse endpoint from db: %w", err)
+		}
+
+		endpoints[idx] = addrPort
+	}
+	node.Endpoints = endpoints
+
+	var hi tailcfg.Hostinfo
+	if err := json.Unmarshal([]byte(node.HostinfoDatabaseField), &hi); err != nil {
+		log.Trace().Err(err).Msgf("Hostinfo content: %s", node.HostinfoDatabaseField)
+
+		return fmt.Errorf("failed to unmarshal Hostinfo from db: %w", err)
+	}
+	node.Hostinfo = &hi
+
+	return nil
+}
+
 func (node *Node) Proto() *v1.Node {
 	nodeProto := &v1.Node{
 		Id:         node.ID,
-		MachineKey: node.MachineKey,
+		MachineKey: node.MachineKey.String(),
 
-		NodeKey:     node.NodeKey,
+		NodeKey:     node.NodeKey.String(),
-		DiscoKey:    node.DiscoKey,
+		DiscoKey:    node.DiscoKey.String(),
 		IpAddresses: node.IPAddresses.StringSlice(),
 		Name:        node.Hostname,
 		GivenName:   node.GivenName,
 		User:        node.User.Proto(),
 		ForcedTags:  node.ForcedTags,
-		Online:      node.IsOnline(),
 
 		// TODO(kradalby): Implement register method enum converter
 		// RegisterMethod: ,
@@ -262,14 +343,17 @@ func (node *Node) Proto() *v1.Node {
 	return nodeProto
 }
 
-// GetHostInfo returns a Hostinfo struct for the node.
-func (node *Node) GetHostInfo() tailcfg.Hostinfo {
-	return tailcfg.Hostinfo(node.HostInfo)
-}
-
 func (node *Node) GetFQDN(dnsConfig *tailcfg.DNSConfig, baseDomain string) (string, error) {
 	var hostname string
 	if dnsConfig != nil && dnsConfig.Proxied { // MagicDNS
+		if node.GivenName == "" {
+			return "", fmt.Errorf("failed to create valid FQDN: %w", ErrNodeHasNoGivenName)
+		}
+
+		if node.User.Name == "" {
+			return "", fmt.Errorf("failed to create valid FQDN: %w", ErrNodeUserHasNoName)
+		}
+
 		hostname = fmt.Sprintf(
 			"%s.%s.%s",
 			node.GivenName,
@@ -278,7 +362,7 @@ func (node *Node) GetFQDN(dnsConfig *tailcfg.DNSConfig, baseDomain string) (stri
 		)
 		if len(hostname) > MaxHostnameLength {
 			return "", fmt.Errorf(
-				"hostname %q is too long it cannot except 255 ASCII chars: %w",
+				"failed to create valid FQDN (%s): %w",
 				hostname,
 				ErrHostnameTooLong,
 			)
@@ -290,49 +374,98 @@ func (node *Node) GetFQDN(dnsConfig *tailcfg.DNSConfig, baseDomain string) (stri
 	return hostname, nil
 }
 
-func (node *Node) MachinePublicKey() (key.MachinePublic, error) {
+// func (node *Node) String() string {
-	var machineKey key.MachinePublic
+// 	return node.Hostname
+// }
 
-	if node.MachineKey != "" {
+// PeerChangeFromMapRequest takes a MapRequest and compares it to the node
-		err := machineKey.UnmarshalText(
+// to produce a PeerChange struct that can be used to updated the node and
-			[]byte(util.MachinePublicKeyEnsurePrefix(node.MachineKey)),
+// inform peers about smaller changes to the node.
-		)
+// When a field is added to this function, remember to also add it to:
-		if err != nil {
+// - node.ApplyPeerChange
-			return key.MachinePublic{}, fmt.Errorf("failed to parse machine public key: %w", err)
+// - logTracePeerChange in poll.go
+func (node *Node) PeerChangeFromMapRequest(req tailcfg.MapRequest) tailcfg.PeerChange {
+	ret := tailcfg.PeerChange{
+		NodeID: tailcfg.NodeID(node.ID),
+	}
+
+	if node.NodeKey.String() != req.NodeKey.String() {
+		ret.Key = &req.NodeKey
+	}
+
+	if node.DiscoKey.String() != req.DiscoKey.String() {
+		ret.DiscoKey = &req.DiscoKey
+	}
+
+	if node.Hostinfo != nil &&
+		node.Hostinfo.NetInfo != nil &&
+		req.Hostinfo != nil &&
+		req.Hostinfo.NetInfo != nil &&
+		node.Hostinfo.NetInfo.PreferredDERP != req.Hostinfo.NetInfo.PreferredDERP {
+		ret.DERPRegion = req.Hostinfo.NetInfo.PreferredDERP
+	}
+
+	if req.Hostinfo != nil && req.Hostinfo.NetInfo != nil {
+		// If there is no stored Hostinfo or NetInfo, use
+		// the new PreferredDERP.
+		if node.Hostinfo == nil {
+			ret.DERPRegion = req.Hostinfo.NetInfo.PreferredDERP
+		} else if node.Hostinfo.NetInfo == nil {
+			ret.DERPRegion = req.Hostinfo.NetInfo.PreferredDERP
+		} else {
+			// If there is a PreferredDERP check if it has changed.
+			if node.Hostinfo.NetInfo.PreferredDERP != req.Hostinfo.NetInfo.PreferredDERP {
+				ret.DERPRegion = req.Hostinfo.NetInfo.PreferredDERP
+			}
 		}
 	}
 
-	return machineKey, nil
+	// TODO(kradalby): Find a good way to compare updates
+	ret.Endpoints = req.Endpoints
+
+	now := time.Now()
+	ret.LastSeen = &now
+
+	return ret
 }
 
-func (node *Node) DiscoPublicKey() (key.DiscoPublic, error) {
+// ApplyPeerChange takes a PeerChange struct and updates the node.
-	var discoKey key.DiscoPublic
+func (node *Node) ApplyPeerChange(change *tailcfg.PeerChange) {
-	if node.DiscoKey != "" {
+	if change.Key != nil {
-		err := discoKey.UnmarshalText(
+		node.NodeKey = *change.Key
-			[]byte(util.DiscoPublicKeyEnsurePrefix(node.DiscoKey)),
+	}
-		)
+
-		if err != nil {
+	if change.DiscoKey != nil {
-			return key.DiscoPublic{}, fmt.Errorf("failed to parse disco public key: %w", err)
+		node.DiscoKey = *change.DiscoKey
+	}
+
+	if change.Online != nil {
+		node.IsOnline = change.Online
+	}
+
+	if change.Endpoints != nil {
+		node.Endpoints = change.Endpoints
+	}
+
+	// This might technically not be useful as we replace
+	// the whole hostinfo blob when it has changed.
+	if change.DERPRegion != 0 {
+		if node.Hostinfo == nil {
+			node.Hostinfo = &tailcfg.Hostinfo{
+				NetInfo: &tailcfg.NetInfo{
+					PreferredDERP: change.DERPRegion,
+				},
+			}
+		} else if node.Hostinfo.NetInfo == nil {
+			node.Hostinfo.NetInfo = &tailcfg.NetInfo{
+				PreferredDERP: change.DERPRegion,
+			}
+		} else {
+			node.Hostinfo.NetInfo.PreferredDERP = change.DERPRegion
 		}
-	} else {
-		discoKey = key.DiscoPublic{}
 	}
 
-	return discoKey, nil
+	node.LastSeen = change.LastSeen
-}
-
-func (node *Node) NodePublicKey() (key.NodePublic, error) {
-	var nodeKey key.NodePublic
-	err := nodeKey.UnmarshalText([]byte(util.NodePublicKeyEnsurePrefix(node.NodeKey)))
-	if err != nil {
-		return key.NodePublic{}, fmt.Errorf("failed to parse node public key: %w", err)
-	}
-
-	return nodeKey, nil
-}
-
-func (node Node) String() string {
-	return node.Hostname
 }
 
 func (nodes Nodes) String() string {