Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add support for scaling kubernetes stateful sets #5

Closed
wants to merge 1 commit into from

Conversation

kormide
Copy link

@kormide kormide commented May 28, 2024

Allow the autoscaler to update stateful sets. This can be useful when workers need to be deployed as a stateful set with persistent volume claims for the FUSE mount's storage.

@EdSchouten
Copy link
Member

Hey! Could you please elaborate a bit more on your use case? There's nothing inherently persistent to workers, so to me it's not clear why you'd want to use stateful sets there. The FUSE mount itself cannot be mounted through a PersistentVolumeClaim. Are you referring to the LocalBlobAccess instance that goes underneath? Why would you want that to be persistent?

@kormide
Copy link
Author

kormide commented May 29, 2024

Hey! Could you please elaborate a bit more on your use case? There's nothing inherently persistent to workers, so to me it's not clear why you'd want to use stateful sets there. The FUSE mount itself cannot be mounted through a PersistentVolumeClaim. Are you referring to the LocalBlobAccess instance that goes underneath? Why would you want that to be persistent?

To maximize performance, each worker is configured to use local node-attached SSDs for the FUSE mount, CAS, filepool, etc. That storage is served by several PVCs for block and filesystem devices which are satisfied from provisioned storage on the node. Each worker pod requires its own set of volumes, hence the need for a stateful set.

I suppose the use of PVCs here is unnecessary and I could mount and partition the SSDs using hostPath since, as you said, there's nothing stateful about a worker 🤔.

@kormide
Copy link
Author

kormide commented May 29, 2024

Yeah, I think I was using the wrong abstraction here. I'll close this for now and I'll reopen if I find a legitimate use case.

@kormide kormide closed this May 29, 2024
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