What server version are you using?

is it known?

That's actually strange since this comes immediately after forcing the task to the stopped state

Would this be it? 

Nope, that's about updating the default container settings back to the task

is it known?

Updating to 1.1.0 gives this error:

ERROR: Could not push back task [e55e0f0ea228407a921e004f0d8f7901] to k8s pending queue [c288c73b8c434a6c8c55ebb709684b28], error: Invalid task status (Task already in requested status): current_status=queued, new_status=queued

Would this be it? https://github.com/allegroai/clearml-agent/commit/6470b16b70165cdab0159ce5613521917e861d79

It's somewhere in the glue/k8s.py file, don't have it in front of me right now 🙂

There was a fix related to this feature on the latest agent version, you'll need to use 1.1.0

If you don’t mind, can you point me at the code where this happens?

I am on 1.0.0

It's part of the k8s glue job to make sure the script will be executed. Are you using the latest agent version v1.1.0?