Description of problem: The alert "The pod disruption budget is below the minimum number allowed pods." is triggerred for every namespace that sets up a virtual machine. Version-Release number of selected component (if applicable): OpenShift version : 4.6.7 kubevirt-hyperconverged-operator : 2.6.1 How reproducible: Always. Steps to Reproduce: 1. Configure an OpenShift cluster, version 4.6.7 2. Enable the virtualization operator 3. Create "HyperConverged" 4. Create VM. Actual results: VM is created successfully. Alert is firing, "The pod disruption budget is below the minimum number allowed pods." # From my namespace, $ > oc get pdb NAME MIN AVAILABLE MAX UNAVAILABLE ALLOWED DISRUPTIONS AGE kubevirt-disruption-budget-d2n2z 2 N/A 0 2d2h $ > oc describe pdb Name: kubevirt-disruption-budget-d2n2z Namespace: xx Min available: 2 Selector: kubevirt.io/created-by=79ad5838-4130-473e-a057-5de3fdda9611 Status: Allowed disruptions: 0 Current: 1 Desired: 2 Total: 1 Events: <none> Expected results: VM is created successfully. No alerts should be firing. Additional info: We can ofcourse silence the alert, eg. like suggested here https://access.redhat.com/solutions/5101781 But that seems like a workaround that shouldn't be necessary, please correct me if I'm wrong. Best regards, Patrik, Sweden
I'm not sure which component this belongs to, I would put it under "Container Native Virtualization (CNV)", but that one doesn't seem to exist anymore, but I guess the team that handles the Virtualization operator would be a good fit. Best regards, Patrik, Sweden
(In reply to Patrik Martinsson from comment #1) > I'm not sure which component this belongs to, I would put it under > "Container Native Virtualization (CNV)", but that one doesn't seem to exist > anymore It's a bit confusing, but CNV is a Product in Bugzilla Moving the bug there. Can you share how you've installed CNV-2.6.1 on OCP-4.6? CNV-2.6 is expected to run on OCP-4.7 and during upgrade on OCP-4.8. On my cluster I see a few $ oc get pdb -A |grep kubevirt|wc -l 3 but certainly not on all namespaces running VMs. $ oc get vmi -A --no-headers | awk '{print $1}'|sort|uniq |wc -l 42 Would you be able to upgrade OCP to 4.7? If the problem remains, can you attach must-gather https://docs.openshift.com/container-platform/4.7/virt/logging_events_monitoring/virt-collecting-virt-data.html#gathering-data-specific-features_virt-collecting-virt-data ? If you are a Red Hat customer, you should probably file a support case at https://access.redhat.com/ and attach this bug to it.
Hi Dan, Thanks for answering. >>Can you share how you've installed CNV-2.6.1 on OCP-4.6? 4.6.7 was a typo, I meant 4.7.6, but yes the installed version of OpenShift Virtualization is 2.6.1 (I installed as usual, through the gui, no problem). $ > oc get pdb -A |grep kubevirt|wc -l 3 $ > oc get vmi -A --no-headers openshift-virtualization-os-images win2k16-functional-lion 3d15h Running 10.144.2.180 tt16.xx.xx openshift-virtualization-os-images win2k16-protective-swallow 4d21h Running 10.144.2.65 tt16.xx.xx So yes, it seems to work as expected, *except* that we get alerts for "The pod disruption budget is below the minimum number allowed pods.". We are a RH customer, but I usually feel that I get better response from Bugzilla's, hence I reported it here. I just don't understand why this alarm is firing, and what is it suppose to do, $ > oc get pdb kubevirt-disruption-budget-d2n2z -o yaml apiVersion: v1 items: - apiVersion: policy/v1beta1 kind: PodDisruptionBudget metadata: creationTimestamp: "2021-04-20T09:56:10Z" generateName: kubevirt-disruption-budget- generation: 1 managedFields: name: kubevirt-disruption-budget-d2n2z namespace: user-manpi ownerReferences: - apiVersion: kubevirt.io/v1alpha3 blockOwnerDeletion: true controller: true kind: VirtualMachineInstance name: vm-manpi-rhel8-3 uid: 79ad5838-4130-473e-a057-5de3fdda9611 resourceVersion: "28026566" spec: minAvailable: 2 selector: matchLabels: kubevirt.io/created-by: 79ad5838-4130-473e-a057-5de3fdda9611 status: currentHealthy: 1 desiredHealthy: 2 disruptionsAllowed: 0 expectedPods: 1 observedGeneration: 1 kind: List metadata: resourceVersion: "" selfLink: "" Any idea about this, or how I can debug this further ? Best, Patrik, Sweden
looks like duplicate to BZ 1873555
Pretty sure this is a duplicate, I'll close this. Thanks for pointing me in the right direction. Best, Patrik, Sweden
*** This bug has been marked as a duplicate of bug 1873555 ***
> We are a RH customer, but I usually feel that I get better response from Bugzilla's, hence I reported it here. What can I say but "the customer is always right". It would be helpful to me if you do file a case and attach it to that other bug. It would help us prioritize the fix and provide better service in general. Regardless, we are interested to hear about how you use OpenShift Virtualization. Please share it with us there.
(In reply to Dan Kenigsberg from comment #8) > > We are a RH customer, but I usually feel that I get better response from Bugzilla's, hence I reported it here. > > What can I say but "the customer is always right". It would be helpful to me > if you do file a case and attach it to that other bug. It would help us > prioritize the fix and provide better service in general. Sure thing, reported it as a support case as well, https://access.redhat.com/support/cases/#/case/02926303 > Regardless, we are > interested to hear about how you use OpenShift Virtualization. Please share > it with us there. Our business simply use the virtualization feature to allow us to migrate older workloads based on .NET Framework X, which we can't rewrite to utilizae .NET Core. So the simplest solution for us is to migrate those VM's (actually create new VM's and just migrate the workload), and take it from there. Doing so we can hopefully take advantage of the other features OCP4x deliveres, such as workflows, cicd, etc. Hope that answers your question somewhat atleast. Best regards, Patrik, Sweden