ReaR does not currently support the backup of empty volumegroups (vgs) in an lvm setup. This means that if a vg is just free space without any lv created, it refuses to take a backup of just free space due to the incapacity of doing the restoration. In RHOSP 17.0 there is one node that has precisely an empty vg called /dev/cinder-volumes. When ReaR tries to verify the disk layout to take the backup of the node, it fails: 2022-04-08 10:14:35.986711968 Verifying that the 'part' entries for /dev/vda in /var/lib/rear/layout/disklayout.conf specify consecutive partitions 2022-04-08 10:14:35.990255874 Verifying that the 'lvm...' entries in /var/lib/rear/layout/disklayout.conf are correct 2022-04-08 10:14:36.028014189 LVM no 'lvmvol /dev/cinder-volumes' for 'lvmdev /dev/cinder-volumes' 2022-04-08 10:14:36.040790096 ERROR: ==================== BUG in /usr/share/rear/layout/save/default/950_verify_disklayout_file.sh line 254: 'Entries in /var/lib/rear/layout/disklayout.conf are broken ('rear recover' would fail)' -------------------- Please report this issue at https://github.com/rear/rear/issues and include the relevant parts from /var/log/rear/rear-controller-0.log preferably with full debug information via 'rear -D mkbackup' ====================
After a conversation with Cinder team, this problem only exists when the deployment type is "lvm", which stores volumes only on one of the controllers. This setup does not offer high-availability or failover capabilities so it is not supported in production setups. It is even highlighted on the Storage documentation, section 2.1 [1]: Red Hat OpenStack Platform supports Red Hat Ceph and NFS as Block Storage (cinder) back ends. By default, the Block Storage service uses an LVM back end as a repository for volumes. While this back end is suitable for test environments, LVM is not supported in production environments. With this information available, as non-production setups are not supported by B&R, the process will ignore the cinder-volumes if it encounters them. This will make the backup and restore process not to fail in case of an "lvm" setup, but at the cost of losing the volumes that might have been stored in the cinder-volumes vg. This is also reasonable because those would entail backuping and restoring workloads, which is also out-of-scope. [1] https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.2/html/storage_guide/assembly-configuring-the-block-storage-service_osp-storage-guide#con-block-storage-back-ends_configuring-cinder
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 (Release of components for Red Hat OpenStack Platform 17.0 (Wallaby)), 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-2022:6543