volcano
volcano copied to clipboard
queue's deserve can not be returned back when other queue's tasks are all not preemptable
What is the problem you're trying to solve
image there are two queues: A and B, and the cluster have 10GPUs, and both A and B have a deserved with 5GPUs and Capability with 10GPUs. Now, user submit jobs with not preemptable tasks to queue A and use up 10GPUs. Then user submit jobs to queue B can not be scheduled.
Describe the solution you'd like
To limit only tasks be preemptable can be scheduled between a queue's deserved and capability in allocate action.
Additional context
No response
Could You elaborate a bit how user can deploy non reclaimable task if queue is reclaimable? I'm just curious.
Hello 👋 Looks like there was no activity on this issue for last 90 days. Do you mind updating us on the status? Is this still reproducible or needed? If yes, just comment on this PR or push a commit. Thanks! 🤗 If there will be no activity for 60 days, this issue will be closed (we can always reopen an issue if we need!).
Closing for now as there was no activity for last 90 days after marked as stale, let us know if you need this to be reopened! 🤗
/reopen
@lowang-bh: Reopened this issue.
In response to this:
/reopen
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.
@lowang-bh As far as I understand issue is with the fact, that non preemtable task should be reclaimable, but it's not. And I will ask again - how do You specify if the task is preemtable? I mean it would be much easier to find proper place in the code which is responsible for skipping non preemtable workloads.
You mean volcano.sh/preemptable: "false" annotation?