villebro opened a new pull request, #2462: URL: https://github.com/apache/datafusion-ballista/pull/2462
## Summary History event logs are collected centrally, so a `JobStart` record should identify which scheduler accepted and planned the job when that information is known. The execution graph model already stores this as optional metadata: normal submitted static/adaptive graphs are constructed with a scheduler id today, but the existing type and comments also allow an execution graph to have no current curator scheduler. Older persisted event logs also predate this field entirely. This change exposes the existing optional graph metadata through the common `ExecutionGraph` interface and writes it into `JobStart` without inventing a scheduler id when one is not available. Fixes #2413. ## Changes - Add optional `scheduler_id` to `JobStart` history events. - Treat `Some(id)` as the recorded scheduler id and `None` as either an older log or an execution graph with no current curator scheduler. - Add `scheduler_id()` to the shared `ExecutionGraph` trait. - Implement `scheduler_id()` for both static and adaptive execution graphs. - Populate `JobStart.scheduler_id` from the graph in scheduler event-log generation. - Remove the stale `dead_code` allowances/comments now that `scheduler_id` is consumed by event logging. - Update tests to cover the serialized scheduler id. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
