Restore oVirt/RHEV VMs and their databases from backup (if any)

This document describes how to restore the VMs running on the oVirt setup.

VMs are backed up everyday with a week retention period.


  1. Gather the UUID of the instance’s disk and the storage domain you are trying to restore from backup.
  2. Logon to one of the oVirt/RHEV hosts.

Go to /mnt/cephfs/ovirt/vmsbackup/

[root@ov1 ~]# ls /mnt/cephfs/ovirt/vmsbackup/
Friday/    Monday/    Saturday/  Sunday/    Thursday/  Tuesday/   Wednesday/

Get the path to the backup from the day you think the VM was running.

  1. df -h to find the mountpoint for the storage domain.                                     3.6T  1.1T  2.4T  32% /rhev/data-center/mnt/                                       7.9T  1.3T  6.7T  16% /rhev/data-center/mnt/glusterSD/
  1. Overwrite the disk for the instance from the backup.

pigz -dc < /mnt/cephfs/path/to/backup > /storage-domain-mntpoint/uuid

chown vdsm:kvm of the disk otherwise it oVirt/RHEV can’t access it.


The databases are backed up at /dbback on the instance itself (e.g., both zabbix instances)

The idea is to delete the existing bad database, reinitialize a blank database, and then execute the sql statements from backup to recreate the database.

In case of mysql databases,

  1. rm -rf /var/lib/mysql/* and restart database.
  2. bzip -dc < /dbback |mysql

For postgresql,

  1. Remove all files except the conf file from /var/lib/pgsql/
  2. Restart and init the database (some postgres commands)
  3. Recreate from backup (Lookup instructions for postgres).