onlyarnav commented on issue #67797: URL: https://github.com/apache/airflow/issues/67797#issuecomment-4606117474
Hey @jason810496 ! Just a heads-up that I'm taking a look at this also because it is similar to #67798 My approach: Similar to what we did in the Java SDK, the Go SDK currently treats every failure as a terminal failed state because it doesn't process the retry context. I am updating messages.go to properly decode should_retry and max_tries from the ti_context, and adding a new RetryTaskMsg type. From there, I'll update executeTask in task_runner.go to emit RetryTaskMsg instead of TaskStateMsg(failed) if a task panics or returns an error while should_retry is true. Question: I noticed the old TODO in task_runner.go specifically called out that we lose retries on exit 0. Just want to confirm—aside from emitting RetryTask, we don't need to change the process exit code itself for the supervisor to handle this correctly, right? -- 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]
