fix: restart unhealthy tailscale proxies before health scan

2026-05-03 04:47:59 +00:00
parent 79f4b95aef
commit 5262c59665
1 changed files with 12 additions and 0 deletions
@@ -1052,6 +1052,18 @@ jobs:
          kubectl annotate storageclass flash-nfs storageclass.kubernetes.io/is-default-class=true --overwrite
          kubectl get storageclass | grep -E "^flash-nfs.*\\(default\\)"
          ! kubectl get storageclass | grep -E "^local-path.*\\(default\\)"
+          tailscale_unhealthy_pods=$(mktemp)
+          kubectl -n tailscale-system get pods -l tailscale.com/managed=true --no-headers \
+            | grep -Ev "[[:space:]](Running|Completed)[[:space:]]" \
+            | awk '{print $1}' >"${tailscale_unhealthy_pods}" || true
+          if [ -s "${tailscale_unhealthy_pods}" ]; then
+            echo "Restarting unhealthy Tailscale-managed proxy pods before final health scan"
+            while read -r pod; do
+              kubectl -n tailscale-system delete pod "${pod}" --wait=false
+            done <"${tailscale_unhealthy_pods}"
+            sleep 30
+            kubectl -n tailscale-system wait --for=condition=Ready pod -l tailscale.com/managed=true --timeout=600s
+          fi
          unhealthy_pods=$(mktemp)
          kubectl get pods -A --no-headers \
            | grep -Ev "[[:space:]](Running|Completed)[[:space:]]" \