-
Notifications
You must be signed in to change notification settings - Fork 111
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
Auto cores #3727
Auto cores #3727
Conversation
max_procs = cluster.job_adapter.nodes.max { |a, b| a.procs <=> b.procs }.procs | ||
maximum = maximum > max_procs ? maximum : max_procs |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Instead of just setting the min and max blindly, we likely need to add directives so it's per cluster. For example at OSC the max for the owens
cluster is 28 and for pitzer it's 48.
Though off the top - I'm not sure if we can achieve this here... Seems like we'd need to apply data-max-auto-cores: 28
to the cluster option itself, not here. So I'm not sure off the top how to solve for that.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we have any existing mechanism for attributes to affect/limit other attributes? If not, maybe setting the max for each cluster here (i.e. data-max-auto-cores-owens
) and choosing which option to show on the front end could work?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we have any existing mechanism for attributes to affect/limit other attributes?
Sure, most things here relate to other things:
Here's auto_qos
populating what accounts it's available for and so on.
hash["data-option-for-auto-accounts-#{account}"] = false |
If not, maybe setting the max for each cluster here
That's the issue. We can't set the max for each cluster here. All the other examples of relationships are between 2 select options. This being a number field - we can't really have the same thing.
I don't think there's anything to do here really. Seems we'll have to add it to add directives to auto_batch_clusters
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yea I would say just set the min to 1 and forgo the max. Then we can fixup auto_batch_clusters
in another PR.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks! I'll create a follow up ticket to update auto_batch_clusters
to set maxes.
Fixes #3680
Sets the minimum number of cores to 1 and the maximum number to the maximum available on a node.
Note: This will not cause the job to be submitted with the specified number of cores until
ood_core
acceptscores
inscript
.