Troubleshooting

This page lists common issues with Jobs and possible solutions, workarounds or debugging tips.

My job has an error

If you see an error on your job on Operator, refer to this table to find out more:

Error name	Link
FAILED_PREPARATION	link
RUNNER_DISPATCH_FAILED	link
RUNNER_FAILED	link
UPLOAD_FAILED	link
NOT_ACCESSIBLE	link

FAILED_PREPARATION

Failed preparation means that there was an internal issue. Retry the job. If you still have this issue after retrying, contact us.

RUNNER_DISPATCH_FAILED

MaaS refused to run the provided workflow. This generally happens if your workflow is invalid. Here are some examples of issues that can cause a RUNNER_DISPATCH_FAILED error:

• Excessive or incoherent RAM/CPU requests. See this section for more details.

• Invalid timeout values

• Invalid file paths, invalid image, invalid mount syntax

• Using both command and commands at the same time.

• Empty workflow (no stages)

• Empty stage (no task in stage)

• Dangerous file mounting behavior

  • Specifically, within the same stage, trying to mount a and a/b in separate tasks. There is a possible attack where a student may try to switch in their own b at just the right time.

• Invalid JSON when using a WORKFLOW document to import the workflow (syntax error or incorrect structure).

• Tried to reference an image that is not allowed in the tenant.

RUNNER_FAILED

This means that the runner has failed during the execution of the workflow. This can happen because:

• The runner tried to pull a Docker image that does not exist

  • Try pulling your images yourself using docker pull

• The pipeline timed out

• A nonzero return code was returned by a task where interruptOnFailure is true

In all cases, you should check the logs of the MaaS pipeline.

UPLOAD_FAILED

Your upload may fail if

• your XML is incorrect (invalid syntax, invalid unescaped special characters like \0)

• the intranet messed up temporarily. As with FAILED_PREPARATION, retry the job and, if the problem still appears, contact us.

NOT_ACCESSIBLE

This means that you have not yet unlocked the submission on which you are trying to submit. This can happen because:

• The graph has changed between the push and the execution of the workflow.
• You are the manager of your project, and you submit a submission without having submitted the previous ones (as a manager, you have read access to the entire graph).