Skip to content

Conversation

@julienmancuso
Copy link
Contributor

What type of PR is this?

/kind documentation

GREP for #270

Signed-off-by: Julien Mancuso <[email protected]>
Signed-off-by: Julien Mancuso <[email protected]>
Copy link
Contributor

@gflarity gflarity left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @julienmancuso!

IMO we should keep the operator focused on Gang Scheduling, Termination and Rolling Updates. It's already very complicated and I'd recommend not adding any additional complexity here.

That said, I'm confident we can achieve the same goals using optionally installable webhooks ( on PodCliques and PodGangs). All we'd need to do is pass on information around what level (PCS, PCSG or PCLQ) via annotations we'd like the compute domains set at. IMO this would be a bit more cloud-native keeping the operator itself focused on what it already does well.

If you have access to a NVL cluster, I'm pretty sure we can bang out a prototype pretty quickly and validate this. Ping me :)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants