aztk
aztk copied to clipboard
Misleading CLI exit codes
We've had some automated jobs silently failing due to exit code 0 being returned in these two error conditions:
- Cluster provisioning with
aztk spark cluster create
failed due to Batch account out of quota - Subsequent attempt to submit a job to that cluster with
aztk spark cluster submit
failed because no master has been selected (or cluster not ready for whatever reason)
Would it be possible to make sure any failed CLI command returns a non-zero exit status in error conditions such as these?
Yeah it seems like when there is an error we just log it but end successfully. We should update that.
My guess is there are more cases where this happens than noted above. We should try to get full coverage here or this will keep coming back.
We are facing similar issues even with 0.10.2 is there any workaround for these issues?