Batch: some jobs fail with no logs or status

[Jorge_blom]

Hi all,

I have a Batch that normally succeeds but sometimes it fails.

When it fails there are no logs and also no statusEvent useful info...

It seems not a quota issue.

Here the relevant lines when cgloud batch jobs describe:

logsPolicy:
  destination: CLOUD_LOGGING
name: projects/{PROJECT_ID}/locations/europe-west6/jobs/{JOB_ID}
state: FAILED
statusEvents:
  - description: Job state is set from QUEUED to FAILED for job projects/{PROJECT_ID}/locations/europe-west6/jobs/{JOB_ID}.
    eventTime: '2023-02-15T07:24:07.877377371Z'
    type: STATUS_CHANGED
taskGroups:
...

 As you can see, from QUEUED to FAILED with no explanation or logs... 

Any help please? (if I restart the same job it works)

[marcanthonyb]

Hi @Jorge_blom ,

Could you please attach the full log of the job that is having the issue. 

Regards,
Marc

[Jorge_blom]

Hi @marcanthonyb there is no logs for this job at all... 😞

[bolianyin]

Hi @Jorge_blom 

In addition to batch_task_logs as log_name by default if you are directed from the Batch job detail page in the console, could you select batch_agent_logs as well to see if you can find anything?

If not, could you provide us the full output from "gcloud batch jobs describe" for us to look further?

[jacksonwb]

In general lack of output from any failures that happen by the batch agent runnables has been a frequent issue for me also. (Like no errors on mispelling a bucket to mount).