goutamadwant commented on issue #12886: URL: https://github.com/apache/seatunnel/issues/12886#issuecomment-6308589176
@nzw921rx I finished a local PoC for this feature . Here’s what I was able to validate: - 2 and 3 masters with `backup-count=1`, recovering from active/standby master, worker and owner-Pod loss without resubmitting the application. - Stable ClusterIP, status/cancel through failover, checkpoint restore and cleanup while preserving user-owned storage. - A real LocalFile pipeline preserved all 480 expected output IDs through worker and active-master loss. All 11 existing single-master tests also pass. i have a few design/scope callouts: - Use an owner/observer Job with fixed master/worker StatefulSets, reusing native Zeta election, backups and recovery. - The validated guarantee is sequential Pod loss with accessible checkpoints. Node/storage outages and partition safety remain unproven. - Keep PostgreSQL fencing separate, and submit the LocalFile restore fix as a separate prerequisite. Does this design and failure-test matrix look right this STIP's follow-up under #11857? if this looks good, i can proceed with the implementation within this scope? -- 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]
