Unable to start elastic PyTorchJob example

[tenzen-y]

Since the base image was changed, it seems now that 'etcd' library is missing. I can no longer run my elastic pytorch jobs that use 'rdzvBackend: etcd'. It works on the previous version, so just wondering if others are experiencing this problem too.

Originally posted by @mathias9395 in #2024 (comment)

[tenzen-y]

@champon1020 Could you investigate the reason why the new image isn't working for elastic PyTorchJob?
As I quickly checked, both the latest and older images don't have a etcdctl.

[champon1020]

OK, I'll investigate it.

[champon1020]

/assign

[tenzen-y]

/assign

Thank you!

[champon1020]

Sorry for the delay in updating progress.

I found that python etcd client library is not included in new image nvcr.io/nvidia/pytorch:24.01-py3 so the pytorch job fails during initializing distributed training.
I'll create a pull request to modify Dockerfile to install python-etcd. (cc: @tenzen-y)