Description of problem: Disk io can't worked if namespace have label mutatevirtualmachines.kubemacpool.io: allocate Version-Release number of selected component (if applicable): 2.4.1 How reproducible: always Steps to Reproduce: 1. create a namespace with label kubemacpool $ oc describe namespace virt-general-disk-io-option Name: virt-general-disk-io-option Labels: mutatevirtualmachines.kubemacpool.io=allocate Annotations: openshift.io/description: openshift.io/display-name: openshift.io/requester: unprivileged-user openshift.io/sa.scc.mcs: s0:c25,c10 openshift.io/sa.scc.supplemental-groups: 1000620000/10000 openshift.io/sa.scc.uid-range: 1000620000/10000 Status: Active No resource quota. No LimitRange resource. 2. create a vm with disk io configuration $cat vm.yaml .... devices: disks: - disk: bus: virtio io: native name: containerdisk .... $oc create -f vm.yaml 3.check vm $oc get vm -o yaml .... - disk: bus: virtio name: containerdisk machine: type: pc-q35-rhel8.2.0 .... Actual results: no disk io configuration Expected results: should have disk io ... - disk: bus: virtio io: native name: containerdisk machine: type: pc-q35-rhel8.2.0 ... Additional info: if remove the label of namespace and create vm, disk io worked well.
Created attachment 1711703 [details] namespace
Created attachment 1711704 [details] vm yaml file
Targeting for 2.5. In 2.4.1, we should keep a release note describing the issue.
@phoracek @ipinto - please complete your review of the PR for this defect: https://github.com/openshift/openshift-docs/pull/24904
SME review was completed for this PR by @phoracek: https://github.com/openshift/openshift-docs/pull/24904 Currently the PR is in peer review.
@dcritch, @phoracek, @ipinto, Please see the PR: https://github.com/openshift/openshift-docs/pull/24904 The peer reviewer has requested this info - why should the io attribute not be used? What is the reason?
I have updated the PR with the workaround noted by Petr, therefore the previous query that I posted above can be ignored. Thank you.
*** Bug 1876475 has been marked as a duplicate of this bug. ***
should be fixed by https://github.com/k8snetworkplumbingwg/kubemacpool/pull/244
Verified in CNV version v2.5.0, with the following scenario; 1. Create a namespace using the attached ns-virt-disk-io.yaml. This namespace spec was extracted from the namespace originally attached to this bug (comment #1), after removing the redundant fields (e.g. managedFields, UIDs, creation timestamps). $ oc apply -f ns-virt-disk-io.yaml 2. Move to the new namespace $ oc project virt-general-disk-io-option 3. Create the VM using the VM spec was attached to this bug in comment #2. 4. Check the created VM spec. $ oc get vm vm-cirros-hpp -o yaml | less spec: running: false template: ... spec: domain: devices: disks: - disk: bus: virtio io: native name: containerdisk Disk IO configuration exists.
Created attachment 1720960 [details] ns-virt-disk-io.yaml