Tolerate ErrDumpInterrupted when listing tun addresses (#1835)

.github/workflows/release.yml

@@ -73,11 +73,8 @@ jobs:
   build-darwin:
     name: Build Universal Darwin
     env:
-      HAS_SIGNING_CREDS: ${{ secrets.APPLE_SIGNING_ROLE_ARN != '' }}
+      HAS_SIGNING_CREDS: ${{ secrets.AC_USERNAME != '' }}
     runs-on: macos-latest
-    permissions:
-      id-token: write
-      contents: read
     steps:
       - uses: actions/checkout@v7
 
@@ -86,68 +83,17 @@ jobs:
           go-version: '1.26'
           check-latest: true
 
-      # GitHub holds ARNs, not credentials, and ARNs outlive a rotation
-      - name: Configure AWS credentials
-        if: env.HAS_SIGNING_CREDS == 'true'
-        uses: aws-actions/configure-aws-credentials@v6
-        with:
-          role-to-assume: ${{ secrets.APPLE_SIGNING_ROLE_ARN }}
-          aws-region: us-east-2
-
-      # parse-json-secrets unpacks into SIGNING_* and ASC_*, masked on the way in
-      - name: Fetch signing credentials
-        if: env.HAS_SIGNING_CREDS == 'true'
-        uses: aws-actions/aws-secretsmanager-get-secrets@v3
-        with:
-          parse-json-secrets: true
-          secret-ids: |
-            SIGNING,${{ secrets.APPLE_SIGNING_DEVELOPER_ID_ARN }}
-            ASC,${{ secrets.APPLE_NOTARY_KEY_ARN }}
-
       - name: Import certificates
         if: env.HAS_SIGNING_CREDS == 'true'
         uses: Apple-Actions/import-codesign-certs@v7
         with:
-          p12-file-base64: ${{ env.SIGNING_P12_BASE64 }}
+          p12-file-base64: ${{ secrets.APPLE_DEVELOPER_CERTIFICATE_P12_BASE64 }}
-          p12-password: ${{ env.SIGNING_PASSWORD }}
+          p12-password: ${{ secrets.APPLE_DEVELOPER_CERTIFICATE_PASSWORD }}
-
-      # The action imports but does not check the chain validates, which is how a p12
-      # missing its intermediate reaches a failing codesign
-      - name: Check the identity is usable
-        if: env.HAS_SIGNING_CREDS == 'true'
-        run: |
-          : "${SIGNING_IDENTITY_SHA1:?empty, so the secret has no identity_sha1}"
-          identities=$(security find-identity -v -p codesigning signing_temp.keychain)
-          case "$identities" in
-          *"$SIGNING_IDENTITY_SHA1"*) ;;
-          *) printf '%s\n' "$identities" >&2; exit 1 ;;
-          esac
-
-      # notarytool wants the key as a file
-      - name: Write the App Store Connect key
-        if: env.HAS_SIGNING_CREDS == 'true'
-        run: |
-          mkdir -p ~/private_keys
-          chmod 700 ~/private_keys
-          key_path="$HOME/private_keys/AuthKey_${ASC_KEY_ID}.p8"
-          (umask 077; printf '%s\n' "$ASC_PRIVATE_KEY" > "$key_path")
-          echo "ASC_P8=$key_path" >> "$GITHUB_ENV"
-
-      - name: Drop the credentials from the environment
-        if: env.HAS_SIGNING_CREDS == 'true'
-        run: |
-          # The action's own inventory, so a new field in a secret is covered
-          python3 -c '
-          import json, os
-          raw = os.environ.get("SECRETS_LIST_CLEAN_UP")
-          if raw is None and os.environ.get("SIGNING_P12_BASE64"):
-              raise SystemExit("SECRETS_LIST_CLEAN_UP is gone, fetched secrets are not being scrubbed")
-          keep = {"SIGNING_IDENTITY_SHA1", "ASC_KEY_ID", "ASC_ISSUER_ID"}
-          names = [n for n in json.loads(raw or "[]") if n not in keep]
-          print("\n".join(f"{n}=" for n in dict.fromkeys(names)))
-          ' >> "$GITHUB_ENV"
 
       - name: Build, sign, and notarize
+        env:
+          AC_USERNAME: ${{ secrets.AC_USERNAME }}
+          AC_PASSWORD: ${{ secrets.AC_PASSWORD }}
         run: |
           rm -rf release
           mkdir release
@@ -156,34 +102,17 @@ jobs:
           lipo -create -output ./release/nebula ./build/darwin-amd64/nebula ./build/darwin-arm64/nebula
           lipo -create -output ./release/nebula-cert ./build/darwin-amd64/nebula-cert ./build/darwin-arm64/nebula-cert
 
-          # Unset in a fork, which has no credentials to sign with
+          if [ -n "$AC_USERNAME" ]; then
-          if [ -n "$SIGNING_IDENTITY_SHA1" ]; then
+            codesign -s "10BC1FDDEB6CE753550156C0669109FAC49E4D1E" -f -v --timestamp --options=runtime -i "net.defined.nebula" ./release/nebula
-            codesign -s "$SIGNING_IDENTITY_SHA1" -f -v --timestamp --options=runtime -i "net.defined.nebula" ./release/nebula
+            codesign -s "10BC1FDDEB6CE753550156C0669109FAC49E4D1E" -f -v --timestamp --options=runtime -i "net.defined.nebula-cert" ./release/nebula-cert
-            codesign -s "$SIGNING_IDENTITY_SHA1" -f -v --timestamp --options=runtime -i "net.defined.nebula-cert" ./release/nebula-cert
           fi
 
           zip -j release/nebula-darwin.zip release/nebula-cert release/nebula
 
-          if [ -n "$ASC_P8" ]; then
+          if [ -n "$AC_USERNAME" ]; then
-            xcrun notarytool submit ./release/nebula-darwin.zip --key "$ASC_P8" --key-id "$ASC_KEY_ID" --issuer "$ASC_ISSUER_ID" --wait
+            xcrun notarytool submit ./release/nebula-darwin.zip --team-id "576H3XS7FP" --apple-id "$AC_USERNAME" --password "$AC_PASSWORD" --wait
           fi
 
-      - name: Drop the signing key
-        if: always() && env.HAS_SIGNING_CREDS == 'true'
-        run: |
-          # Locked, not deleted: import-codesign-certs deletes it in its own post
-          # step and fails the job if it is already gone. Locked is unusable.
-          security lock-keychain signing_temp.keychain || true
-          rm -f "$ASC_P8"
-          # Nothing later in this job needs AWS
-          python3 -c '
-          import json, os
-          names = json.loads(os.environ.get("SECRETS_LIST_CLEAN_UP") or "[]")
-          names += ["ASC_P8", "SIGNING_IDENTITY_SHA1", "ASC_KEY_ID", "ASC_ISSUER_ID",
-                    "AWS_ACCESS_KEY_ID", "AWS_SECRET_ACCESS_KEY", "AWS_SESSION_TOKEN"]
-          print("\n".join(f"{n}=" for n in dict.fromkeys(names)))
-          ' >> "$GITHUB_ENV"
-
       - name: Upload artifacts
         uses: actions/upload-artifact@v7
         with:

overlay/tun_linux.go

@@ -5,6 +5,7 @@ package overlay
 
 import (
 	"encoding/binary"
+	"errors"
 	"fmt"
 	"io"
 	"log/slog"
@@ -483,8 +484,17 @@ func (t *tun) addIPs(link netlink.Link) error {
 	//iterate over remainder, remove whoever shouldn't be there
 	al, err := netlink.AddrList(link, netlink.FAMILY_ALL)
 	if err != nil {
+		//RTM_GETADDR dumps the whole system, so any concurrent address change
+		//interrupts it - including the kernel's async tentative->preferred
+		//flip of an IPv6 address the AddrReplace calls above just added,
+		//which makes this a race against our own setup. Partial results are
+		//still returned; the worst case is a stale address surviving until
+		//the next config reload, which beats failing startup over it.
+		if !errors.Is(err, netlink.ErrDumpInterrupted) {
 			return fmt.Errorf("failed to get tun address list: %s", err)
 		}
+		t.l.Warn("tun address list dump was interrupted, stale addresses may remain")
+	}
 
 	for i := range al {
 		if hasNetlinkAddr(newAddrs, al[i]) {