This issue is to request an ACM 2.9 release note be added to describe bug MGMT-16119 and its work around. MGMT-16119 fix will not make it into ACM 2.9.

Issue: MGMT-16119 is related to ACM DR backup / restore functionality. Baremetal spoke clusters that are deployed via Infrastructure Operator using the ZTP flow are impacted by this bug. If the resources for this type of cluster are backed up and restored to a secondary hub cluster using the ACM backup/restore feature, the spoke cluster will perform a re-installation onto its nodes, which will destroy the existing spoke cluster. This only affects baremetal spoke clusters which were deployed using zero touch provisioning, meaning that they have BareMetalHost resources that manage powering on and off baremetal nodes, as well as attaching virtual media for booting.

If a BMH resource was not used in the deployment of the spoke cluster, there will be no impact upon DR restoral of the spoke.

Work Around: The workaround involves excluding spoke BMH resources on the primary hub from being backed up and restored to the secondary hub. This can be achieved by adding the following label to the BMH resources on the primary hub: 

      velero.io/exclude-from-backup: "true"

This label is used to exclude any resource from being backed up. 

The impact of excluding the BMH resource from restoral is that removing a cluster via zero touch provisioning will not 100% function as the BMH manages power for the baremetal nodes.