ebcfee5f20
The backup&restore process on the controller needs to be done with the controller stopped. If that condition is not met, we risk introducing race conditions that could create unrestorable non-consistent backups. Currently, the B&R process stops the three controllers to perform a backup, leaving the cloud unmanageable during the duration of the backup, which can take several minutes. This change introduces the capability of doing "serial" backups. Managing Pacemaker lets us take a node out of the controller Pacemaker cluster, take a backup from it and introduce it again on the cluster. While that backup is being done, the other two nodes of the controller cluster continue to be available to manage the cloud. Bugzilla: #1877798 Change-Id: I77c4e6accc205cbc962c663babe1018b51d8d266 |
||
|---|---|---|
| .. | ||
| defaults | ||
| meta | ||
| molecule/default | ||
| tasks | ||
| templates | ||
| vars | ||