Fix disk usage bug #4473
Fix disk usage bug #4473
Conversation
| } | ||
| if self.get_bundle_state(bundle.uuid) != State.FAILED: |
There was a problem hiding this comment.
Note: This was another bug that occurred, but this time due to the PR that tracks disk usage while a bundle is running. Basically, in bundles.py, the bundle is failed during the UPLOADING_RESULTS phase here if the user doesn't have enough disk quota to upload the results. When this would happen, the worker isn't updated on the failure status, so it just goes and hits the bundle checkin and finalizes the bundle, and in transition_bundle_finished the bundle is marked as FAILED because it has a failure_message in its metdata from the code in bundles.py. But, when that bundle checkin occurs, the data size is updated again which is WRONG because the actual size of the bundle ON THE SERVER (or Azure blob) is zero even though on the worker it still has a real size (until it gets deleted later)
…e errors elsewhere for kill and run
Fix for #4472. This just reverts to using disk computation from scratch while I look for the issue with the prior change.