SEZ9 commented on PR #12710: URL: https://github.com/apache/seatunnel/pull/12710#issuecomment-6356517921
Following up on the three points from my earlier pass, against the current head `016f3d4c` (`016f3d4c89d5a1857d1082b694590f2bb1e65ce9`): **Grace period on the master/worker Deployments (F1/F2):** this is resolved on the chart side. `deploy/kubernetes/seatunnel/values.yaml` sets `terminationGracePeriodSeconds: 30` for both master and worker, and `deployment-seatunnel-master.yaml` / `deployment-seatunnel-worker.yaml` render it, so the default is not opt-in and the JVM now gets a real window to run the Hazelcast shutdown hook before SIGKILL. The remaining half of the finding, that `SeaTunnelServer.shutdown` ignores the terminate flag so the resulting stop is Hazelcast `TERMINATE` rather than a clean job stop, is an engine-side question and I'm fine deferring it out of this PR. I am not treating it as fixed by the chart settings, though, so please open the follow-up issue and link it in the PR description so the deferral has an owner and traceable scope. **Docs (F3):** the PR now carries matching en/zh updates for the Kubernetes graceful-shutdown behaviour. Please just confirm that the pages no longer present the `preStop` + `stop-seatunnel-cluster.sh` workaround as the way to get a clean stop, and instead describe the `exec`-based SIGTERM path plus `terminationGracePeriodSeconds` as the primary mechanism. Nothing new from the `dev` merge touches the PR-owned files beyond the unrelated test addition in `SeaTunnelClusterScriptTest.java`, so my source-level conclusion stands. Concrete remaining asks before I approve: 1. Link the follow-up issue for the `SeaTunnelServer.shutdown` terminate-flag question in the PR description. 2. Get a green Build on this exact head (or the updated head if you push the description-only change). If the fault-tolerance lane is the only failure, rerun that job rather than adding unrelated engine changes to chase it. Thanks for keeping the branch current. <!-- streview-comment:1968 --> -- 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]
