orangeCatDeveloper commented on issue #3849:
URL: https://github.com/apache/hertzbeat/issues/3849#issuecomment-5231140397
Reproduced this on a kind cluster with chart 1.7.3 — the root cause is in
the helm chart:
**The bare `livenessProbe` kills the manager mid-boot.** The chart defines
`livenessProbe: tcpSocket: 1157` with no `initialDelaySeconds` and no
`startupProbe`, so with k8s defaults the container is killed ~30s after start.
The manager typically needs well over 30s to boot, so every attempt is killed
halfway and the pod crashloops forever:
```
Warning Unhealthy kubelet Liveness probe failed: dial tcp
10.244.0.36:1157: connect: connection refused
Warning BackOff kubelet Back-off restarting failed container hertzbeat
```
**Fix:** apache/hertzbeat-helm-chart#23 adds 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 — permanent
CrashLoopBackOff; after — all pods reach `1/1 Running`.
**Workaround for existing 1.7.3 installs** until a new chart release:
```bash
kubectl patch deployment hertzbeat -n monitoring --type json -p '[
{"op":"add","path":"/spec/template/spec/containers/0/startupProbe",
"value":{"tcpSocket":{"port":1157},"periodSeconds":10,"failureThreshold":30}}]'
```
@tomsun28 could you take a look at apache/hertzbeat-helm-chart#23 when you
have a chance? Thanks!
--
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]
