orangeCatDeveloper opened a new pull request, #23: URL: https://github.com/apache/hertzbeat-helm-chart/pull/23
## What's changed? Fixes apache/hertzbeat#3849: `helm install` leaves the manager pod in permanent CrashLoopBackOff on many clusters. Two compounding causes, reproduced on a kind cluster with chart 1.7.3: 1. The manager starts before postgres is ready, fails with `Connection to hertzbeat-database:5432 refused` and exits. This part self-heals via restarts once the database is up. 2. The bare `livenessProbe` (tcpSocket 1157, no `initialDelaySeconds`, no `startupProbe`) kills the container ~30s after start with k8s defaults, while the manager needs well over 30s to boot on typical nodes. Every retry gets killed mid-boot, so the pod never recovers — `kubectl describe` shows `Liveness probe failed: dial tcp ...:1157: connect: connection refused`. The crash is invisible in `kubectl logs` because logback routes everything to file appenders inside the container, which is why the issue reports "no useful log". Fix: add a `startupProbe` (10s × 30 = 5 min boot budget) to the manager and collector deployments; liveness only takes over after startup succeeds. Verified on kind: before — manager restarts forever; after — one restart while postgres boots, then all pods reach `1/1 Running`. ## Checklist - [x] I have read the [Contributing Guide](https://hertzbeat.apache.org/docs/community/code_style_and_quality_guide) - [x] I have written the necessary doc or comment. - [ ] I have added the necessary unit tests and all cases have passed. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
