Rollback on error

Note:

Reverting to the previous version may be needed in the following scenarios:

• There is failure during upgrade, which is beyond recovery;
• There is an accidental data loss or corruption;
• Upgraded cluster is unusable.

To roll back to a previous version, take the following steps:

1. Make a copy of cluster_config.json.
2. Verify that the etcd backup data is present on the primary server at the following location: /mnt/backup/backup/<etcdBackupPath>/<node-name>/snapshots.
   • etcdBackupPath – the same as the one specified in backup.json while enabling the backup.
   • node-name – the hostname of the server node.
3. To check the backupstore volume backup directory, run the following command:

   ls /mnt/backup/backupstore

4. Uninstall AI Center on all the nodes.
5. Now restore to the previous version of AI Center. Make sure to follow the Backing up and restoring the cluster documentation available for that specific version.
6. Once the restore is complete, make sure you disable the maintenance mode:

   /path/to/old-installer/configureUiPathAS.sh disable-maintenance-mode

7. Verify that maintenance mode is disabled by executing the following command:

   /path/to/old-installer/configureUiPathAS.sh is-maintenance-enabled

8. During the upgrade, scheduled backups were disabled on the primary node. To enable them again, run the following command:

   /path/to/old-installer/configureUiPathAS.sh resume-scheduled-backups