Prerequisites for MongoDB restore and recovery
Review the following prerequisites and limitations before you begin the restore or the recovery process:
Ensure that your source MongoDB cluster and the target MongoDB cluster have the same:
MongoDB version
Authentication type
If the target MongoDB cluster runs under the
mongodaccount, you need a non-rootsudoeraccount. Also configure it as a host user intpconfigbefore you start the restore. NetBackup recovers and runs the instance under the target cluster's user.(Restoring MongoDB data to an alternate cluster) Ensure that you add the alternate MongoDB cluster credentials in the source cluster credentials file.
Ensure that the PEM files or security certificates are available on the destination cluster before you start the recovery operation.
The authentication type of the target cluster during the recovery process must be the same as the authentication type during the backup.
Ensure that the target MongoDB cluster has sufficient free storage space for restoring the data.
Select only one full backup image group and its appropriate incremental images. If you select more than one full backup image group, the recovery may fail as the restored data can get corrupted.
NetBackup for MongoDB plug-in does not support cross-platform file system restore. For example, XFS to ext4 or vice versa is not supported.
Ensure that the HostUser value that is defined in the tpconfig command is the same as the host user account that is used to configure the MongoDB cluster. (MongoDB daemon's host user account.)
For the destination client, select a backup host before you submit the restore job.
Point-in-time recovery is valid only for recovery from incremental backups.
Canceling a parent job in a compound restore job does not cancel the child restore jobs. You must manually cancel the child restore jobs as well.
After you start a restore or a recovery job in the web UI, look for the job record and status in the Task progress section in the node. The job might take some time to appear in the list and the compound job can take time to start the parent pre-recovery check. Click to refresh the task list.
In a Restore Only operation for a sharded cluster, follow the standard Restore Only steps:
Shut down all the MongoDB processes (mongos or mongod) before starting the restore.
The
tpconfighost user must have permission to the target folder in the original restore and alternate restore.The MongoDB log file paths remain the same from the original configuration. If you do an alternate restore:
Make sure that the same path is available during restore.
Change the log file path in the configuration file for mongod or mongos process after a successful recovery.
The path to the
.pidfiles must be available on the destination MongoDB cluster to ensure that the recovery operation is successful.Special consideration is needed when you back up multiple MongoDB clusters that run on the same server. (Backups that use the same or different backup policies.) Ensure that you select the correct application server that you want to restore.
For example, if there are multiple clusters with the following configuration:
Replica1 Primary: host1:26050 Secondary: host1:26060 Replica2 Primary: host1:26055 Secondary: host1:26066
And you want to recover Replica1, ensure that you specify the correct application server and its port (host1-26050) as the .
Before you start a recovery operation, ensure that stale mdbserver (thin client) processes are not running on your MongoDB instance at the following path:
/<mdbserver_location>/<Host>-<MongodPort>-<mdbserver_port in range>/mdbserver
If any of these stale processes are running, the recovery operation is unable to shut down the MongoDB instance which causes the recovery job to stop responding.
Restore and recovery of the MongoDB cluster requires the same security mode that is used at the time of the backup. Ensure that the security mode is the same for the original cluster and the target cluster.
For example, if SSL is used during the backup, then recovery is done using SSL and the target configuration is changed to SSL. Similarly, if TLS is used during the backup, then recovery is done using TLS and the target configuration is changed to TLS.
Restore and recovery of the MongoDB cluster requires the same value of the Feature Compatibility Version (FCV) that is used at the time of the backup. Ensure that FCV is the same for the original cluster and the target cluster.
For example, if FCV is 4.2 during the backup, then restore uses FCV 4.2 and the target cluster has FCV 4.2 after the recovery process completes. Similarly, if FCV is 4.0 during the backup, then restore uses FCV 4.0 and the target cluster has FCV 4.0 after the recovery process completes.