[NodeResourceTopology] filter rejects pods using ephemeral-storage
philsphicas opened this issue · comments
Area
- Scheduler
- Controller
- Helm Chart
- Documents
Other components
No response
What happened?
Scheduling of pod that should be NUMA aligned but uses ephemeral storage fails with "cannot align pod".
Type Reason Age From Message
---- ------ ---- ---- -------
Warning FailedScheduling 74s default-scheduler 0/9 nodes are available: 2 Insufficient hugepages-1Gi, 4 Insufficient cpu, 5 cannot align pod: test-deployment-55fb46fb77-2dz5t. preemption: 0/9 nodes are available: 9 No preemption victims found for incoming pod.
The filter logic rejects it because ephemeral-storage lacks NUMA affinity.
I0905 20:17:32.330205 1 filter.go:141] "early verdict" logID="default/test-deployment-55fb46fb77-2dz5t/test-deployment-1-container-1" node="b37dev08a1co03" resource="ephemeral-storage" suitable="false"
Special handling may be required, similar to #355
What did you expect to happen?
The filter should ignore ephemeral-storage when considering NUMA alignment.
How can we reproduce it (as minimally and precisely as possible)?
No response
Anything else we need to know?
No response
Kubernetes version
$ kubectl version
Client Version: version.Info{Major:"1", Minor:"26", GitVersion:"v1.26.3", GitCommit:"f18584a06fc476806da2e340e2eed960659871e8", GitTreeState:"clean", BuildDate:"2023-06-12T18:45:20Z", GoVersion:"go1.19.7", Compiler:"gc", Platform:"linux/amd64"}
Kustomize Version: v4.5.7
Server Version: version.Info{Major:"1", Minor:"26", GitVersion:"v1.26.3", GitCommit:"f18584a06fc476806da2e340e2eed960659871e8", GitTreeState:"clean", BuildDate:"2023-06-12T18:45:20Z", GoVersion:"go1.19.7", Compiler:"gc", Platform:"linux/amd64"}
Scheduler Plugins version
v0.26.7