-
Notifications
You must be signed in to change notification settings - Fork 251
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 AssumedWorkloads
to cache.ClusterQueue
#267
Comments
/assign |
assumedWorkloads
's struct in cache to map[string]sets.StringassumedWorkloads
's struct in cache to map[string]sets.String
assumedWorkloads
's struct in cache to map[string]sets.StringAssumedWorkloads
to cache.ClusterQueue
Why did we need to know if a workload was assumed? cc @ahg-g who reviewed the PR |
We hold the deletion until the admitted workloads finished, for unadmitted workloads, they may never have change to run e.g. insufficient resources, if so, clusterQueue will stuck in terminating. But the problem left here is these workloads are isolated, but we can collect them again if we create a same name clusterQueue. |
Not sure my question is answered. IIUC, you are saying that when an assumed workload gets forgotten, it doesn't change the status of the workload, then we wouldn't reconcile the clusterQueue again to remove the finalizer. How does the PR for this issue solved that? I think we can use a generic event for when a workload is forgotten to reconcile the clusterqueue |
I am not following the reasoning as well, but the highlevel semantics we are looking for is described in https://github.com/kubernetes-sigs/kueue/pull/284/files#r902946904 |
Let me explain this more clearly:
|
lets have this discussion on the issue. |
Revert the commit #286 after talking with ahg-g |
What would you like to be added:
Currently
cache.assumedWorkloads
contains the whole admitted workloads, however, we can't easily tell whether a clusterQueue holds admitted workloads or not from this struct. This is useful in reconciling for clusterQueue holding admitted workloads should not be deleted directly. #134 (comment)So I'd like to add a new field
AssumeWorkloads
toClusterQueue
:When we assume a workload, we'll insert the workload name to
AssumedWorkloads
, when we delete/forget workload, we'll remove it.Why is this needed:
Completion requirements:
This enhancement requires the following artifacts:
The artifacts should be linked in subsequent comments.
The text was updated successfully, but these errors were encountered: