SEZ9 commented on PR #12710: URL: https://github.com/apache/seatunnel/pull/12710#issuecomment-6116509875
@ricky2129 thanks for the update, and sorry for the delayed answer. On what is still open from my earlier round: - `terminationGracePeriodSeconds` on both Deployments plus the `values.yaml` default is now in on `4501d089`, which covers the grace-period half of the master/worker findings. What I still want to settle is the shutdown policy: with no `hazelcast.shutdownhook.policy` set, the hook runs `TERMINATE`, and `SeaTunnelServer.shutdown` does not look at the terminate flag, so the pod leaves quickly but it is an abrupt teardown rather than the clean stop the PR description suggests. Either wire a graceful policy (and make `SeaTunnelServer.shutdown` honour it) or, if you prefer to keep that out of this PR, please adjust the description and docs so they describe the behaviour as a fast, clean exit from the cluster rather than a graceful job stop. - Docs: the new tip block in the hybrid/separated K8s pages is good. Could you check whether the graceful-shutdown section that presents `preStop` + `stop-seatunnel-cluster.sh` as the way to get a clean stop still reads as the recommended path? Now that SIGTERM reaches the JVM, it should be positioned as optional, with the exec-form `command` and `terminationGracePeriodSeconds` as the primary mechanism. Once those two are resolved (or explicitly deferred with the wording adjusted), I'm happy to approve. <!-- streview-comment:1683 --> -- 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]
