Job failed even with "CircleCI received exit code 0"


I have a small job defined like this

          command: |
            export CONNECT_TO_KUBE=1
            python3 -m training.kube.<< parameters.script >>

The step on CircleCI show code wrapped with bash an proper “piping” and attributes

#!/bin/bash -eo pipefail
python3 -m training.kube.pipelines

The step failed but output show exit status 0

2024-01-11 15:54:45,832 ::  training.kube.create :: INFO :: Creating new experiment for master...
2024-01-11 15:54:45,860 ::  training.kube.create :: INFO :: Experiment already exists for master, doing nothing!
CircleCI received exit code 0

I think the answer is here

I am gonna make a test.

It doesn’t change anything. I opened a support ticket.

One question.

What status code do you expect kube.create to return in this situation?

The reason for asking is that the log shows that kube.create is just treating the issue as an INFO level issue when logging, which for most systems would not be a cause of a non-zero error status.

The status is correct. The thing I do not understand is why the job is marked as failed.