Commit Graph

1 Commits

Author SHA1 Message Date
Matt Aitken 0591db5f2b Fix for short runs inside (batch)triggerAndWait (#1263)
* Test for checkpoints

* Make sourceTaskAttemptId optional on resumeBatchRun

* Removed all completions/executions logic from the shared queue consumer

* Removed the sourceTaskAttemptId from ResumeBatchRunService

* Revert "Removed all completions/executions logic from the shared queue consumer"

This reverts commit d35398d50463c81a1975bb5d5bcfca66a24b8ede.

* WIP on triggerAndWait…

* Fixed triggerAndWait continuing when a checkpoint completes

* Remove the ResumeAttempt code that fails attempts (was protecting against infinite restores)

* Removed messageBody.data.completedAttemptIds.length === 0 commented out code

* Don’t ack if there’s no batchRun

* Added the marqs?.replaceMessage back in but NOT when there’s no checkpoint. More logging

This is a fix for when some attempts fail

* Improvement to the test task that now randomly fails attempts

* When a checkpoint happens, only continue the attempt if it’s in the correct state

* Changeset for rollback in branch

* Set keepRunAlive to false when the dependent task isn’t finished

* Changeset manual version (to get inline with the hotfix branch)

* Changeset: Fixes for continuing after waits

* Latest lockfile (after manual changeset version)
2024-08-20 13:57:08 +01:00