This alert fires when integrated Node Maintenance Operator (NMO) custom resources (CRs) are detected. This alert only affects OKD 1.6.
The presence of NodeMaintenance CRs belonging to the
nodemaintenance.kubevirt.io API group indicates that the node specified in
spec.nodeName was put into maintenance mode. The target node has been
cordoned off
and drained.
You cannot upgrade to OKD 1.7.
Check the kubevirt-hyperconverged resource to see whether it is upgradeable:
$ kubectl get hco kubevirt-hyperconverged -o json | jq -r '.status.conditions[] | select(.type == "Upgradeable")'
Example output:
{
"lastTransitionTime": "2022-05-26T09:23:21Z",
"message": "NMO custom resources have been found",
"reason": "UpgradeBlocked",
"status": "False",
"type": "Upgradeable"
}
Check for a ClusterServiceVersion (CSV) warning event such as the following:
Warning NotUpgradeable 2m12s (x5 over 2m50s) kubevirt-hyperconvergedNode
Maintenance Operator custom resources nodemaintenances.nodemaintenance.kubevirt.io
have been found.
Please remove them to allow upgrade. You can use NMO standalone operator if
keeping the node(s) under maintenance is still required.
Check for NMO CRs belonging to the nodemaintenance.kubevirt.io API group:
$ kubectl get nodemaintenances.nodemaintenance.kubevirt.io
Example output:
NAME AGE
nodemaintenance-test 5m33s
Remove all NMO CRs belonging to the
nodemaintenance.nodemaintenance.kubevirt.io/ API group. After the integrated
NMO resources are removed, the alert is cleared and you can upgrade.
If a node must remain in maintenance mode during upgrade, install the Node
Maintenance Operator from OperatorHub. Then, create an NMO CR belonging to the
nodemaintenance.nodemaintenance.medik8s.io/v1beta1 API group and version for
the node.
See the HCO cluster configuration documentation for more information.
If you cannot resolve the issue, see the following resources: