ricky2129 commented on PR #12710: URL: https://github.com/apache/seatunnel/pull/12710#issuecomment-6118641423
Thanks @SEZ9 and @DanielLeens, really appreciate you both going through this so carefully. On the shutdown policy, I'd like to keep that out of this PR. Getting `SeaTunnelServer.shutdown` to honour a graceful policy changes how running jobs stop, and I think it deserves its own discussion and testing. So I went with adjusting the wording instead. The description now says plainly that this gives a fast, clean exit from the cluster on SIGTERM, not a graceful stop of running jobs. The docs also mention that `GRACEFUL` only lets Hazelcast migrate partition data before the member leaves, and that tasks still recover from their last checkpoint. For the docs, you were right that the old `preStop` snippet still read like the main way to stop cleanly. In the hybrid and separated K8s pages (en/zh) the section is now "Health Check and Shutdown". It leads with the exec-form `command` and `terminationGracePeriodSeconds`, which the StatefulSets on those pages already use. `preStop` + `stop-seatunnel-cluster.sh` is now an optional tip for setups where the command has to stay `/bin/sh -c`. The stop script only sends the same SIGTERM to the `SeaTunnelServer` process anyway. I also updated the "Rolling Updates" list in `operations.md` and added a short note with a link in `tuning-guide.md` 7.4, both in en/zh. I picked up the two small optional ones too: `values.yaml` now notes that `0` is treated as unset, and there's a one-line comment above `exec java` in the script. The last CI failure was the `CassandraIT` flake on Flink 1.20. This push starts a fresh run, and I'll share the link if anything unrelated fails again. The readiness/startup probes and the optional PDB will come in a separate chart PR. -- 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]
