Fedora Account System
Red Hat Associate
Red Hat Customer
Description of problem: On CNV v4.10.2, if you want to upgrade to 4.11 with k8s-nmstate, you must install the k8s-nmstate standalone operator (KNO). After installing it, any change of placementConfiguration on HCO (and hence - on CNAO) is still applied on KNO. Any attempt to set placementConfiguration on KNO is overriden by CNAO. Version-Release number of selected component (if applicable): CNV v4.10.2 CNAO v4.10.2-3 HCO v4.10.2-2 nmstate 1.2.1-2.el8_6.x86_64 How reproducible: Always Steps to Reproduce: 1. Install KNO (if not already installed). I am using the internal script https://gitlab.cee.redhat.com/contra/cnv-qe-automation/-/blob/master/nmstate/deploy.sh for that (don't forget to export CLUSTER_NAME and CLUSTER_DOMAIN first). 2. Edit HCO $ oc edit hco kubevirt-hyperconverged -n openshift-cnv 3.Apply placementConfiguration on HCO, for example: spec: ... workloads: nodePlacement: nodeSelector: kubernetes.io/hostname: <some cluster worker name> 4. Wait for the edited HCO to be applied, for example - by waiting for all pods in openshift-cnv namespace to be in "Running" state. $ oc get pods -n openshift-cnv -w 5. Look for nmstate-handler pods -n openshift-nmstate namespace $ oc get pods -n openshift-nmstate -o wide | grep handler Actual results: One pod only, on the node used for HCO placementConfiguration nmstate-handler-zhgtz 1/1 Running 0 6h36m 192.168.0.80 n-yoss-411-nnwnt-worker-0-sv22v <none> <none> Expected results: Should be an nmstate-handler pod for each cluster node nmstate-handler-24sqm 1/1 Running 0 28s 192.168.3.130 n-yoss-411-nnwnt-master-2 <none> <none> nmstate-handler-94gr2 1/1 Running 0 28s 192.168.2.193 n-yoss-411-nnwnt-master-1 <none> <none> nmstate-handler-hjdjv 1/1 Running 0 28s 192.168.0.155 n-yoss-411-nnwnt-worker-0-fsh8z <none> <none> nmstate-handler-rbdm7 1/1 Running 0 28s 192.168.3.199 n-yoss-411-nnwnt-worker-0-vklgn <none> <none> nmstate-handler-rqnjh 1/1 Running 0 28s 192.168.1.219 n-yoss-411-nnwnt-master-0 <none> <none> nmstate-handler-zf5z2 1/1 Running 0 11s 192.168.0.80 n-yoss-411-nnwnt-worker-0-sv22v <none> <none> Additional info: My assumption is that this is a CNAO bug, therefore opened for CNV Network.
Verified on CNV 4.10.3 OCP 4.11.0-0.nightly-2022-06-06-201913 cluster-network-addons-operator v4.10.3-2 Verified by repeating the same scenario as in the bug description, and getting the expected result (nmstate-handler pods in openshift-nmstate namespace exist on every worker node).
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (OpenShift Virtualization 4.10.3 Images), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHEA-2022:5675