Troubleshooting Rancher upgrades

Upgrading to Rancher v2.7.2 will fail if SUSE® Rancher Prime OS Manager clusters are defined. The rancher pod gets stuck in a crash loop (see https://github.com/rancher/rancher/issues/41145).

Note that the issue is present only if at least one SUSE® Rancher Prime OS Manager cluster is defined.

To workaround the issue create an empty dynamicschemas.management.cattle.io resource named machineinventoryselectortemplate:

kubectl apply -f - <<EOF
kind: DynamicSchema
apiVersion: management.cattle.io/v3
metadata:
  name: machineinventoryselectortemplate
EOF

The crash loop will stop and Rancher upgrade will complete successfully.