pod terminate after command finishes
Hi folks -- it seems like if runpod notices that the entrypoint command for my pod finishes, it restarts the container and runs it again. is that expected, and is there any way to turn that off and have the pod terminate instead of re-running?
6 Replies
It should be that way In docker
So if you don't want it to stop then run an application
i do actually want it to stop -- it's a training run for a model, and the docker command that runs does one training round, and it seems like the container just re-starts when the command exits with status 0
i am just using the runpod pytorch base container but passing command args in my runpod template -- anyone know at what level the restart might be happening?
it'd be great to have a "clean up on terminate" option as well so i dont need to log into the dashboard and clean up instances that stop once completed
entrypoint command is something that overwrites RunPod settings. What I would do is have script that starts training and after it’s done call runpodctl to stop pod
that makes sense! if i could log a feature request -- it'd be great to not have to do that all manually myself, seems like a great checkbox to have right in the prodcut
Oh sorry i read that wrong
no stress! stopping the pod from within it worked fine 👍