DanielLeens commented on issue #11667: URL: https://github.com/apache/seatunnel/issues/11667#issuecomment-5229109945
This reads like a useful Zeta observability feature rather than a bug. The important part is that the proposal is not just "show the latest error better". It is really asking for a stable failure-history contract: multiple failure entries, attempt grouping, task attribution, timestamps, and retention behavior for finished jobs. That is why treating it as a dedicated design-first feature makes sense. The strongest part of the issue is that most of the raw ingredients already exist in some form today: - task and sub-plan failures are already observed, - state-transition timestamps are already recorded, - the gap is mainly that only the first error is retained and none of that history is exposed as a structured read path. Before implementation starts, the key things to pin down are: 1. how many failure-history entries are retained and for how long; 2. whether attempt/restart grouping becomes part of a stable public contract; 3. how this interacts with finished-job history storage so we do not create a feature that only works reliably for running jobs. The direction itself looks reasonable to keep open and continue refining. Since this is already labeled `help wanted`, contributions are very welcome once the design boundary is explicit enough. -- 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]
