DanielLeens commented on issue #11735: URL: https://github.com/apache/seatunnel/issues/11735#issuecomment-5794573475
Classification: D / Zeta bounded task-failure-history design. The revised STIP resolves the important compatibility direction: current `dev` increments `SubPlan.pipelineRestoreNum` as part of restore scheduling, so keeping a separate durable diagnostic attempt identity is the right way to avoid changing the retry budget. The proposed `pipelineId + attempt + taskGroupId` first-write-wins key, explicit truncation flags, bounded text limits, terminal-time-anchored retention, and no worker-address attribution are also now concrete enough to review as v1. Keep #11734 as a draft design until the implementation contract proves three remaining boundaries: the diagnostic-attempt advance must be fenced/idempotent across active-master change without altering retry eligibility; asynchronous best-effort failure writes must not block or re-enter the failure/restore path; and `TaskExecutionState` wire compatibility must be preserved explicitly before optional structured failure fields are added. Tests must cover those boundaries, late writes after terminal cleanup, and a known job with no retained records versus an unknown/expired job at the REST boundary. No assignment or label change is made in this batch. Once those details are frozen in #11734, the implementation can be split into small, independently reviewable slices. -- 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]
