We have been considering using Percona backup for MongoDB (PBM). I can see it has a lot of advantages, but no selective (e.g. collection level restores). This missing selective restore option could be a dealbreaker, so I try to come up with a solution to work around that.
Does anyone have information about the way a backup is stored and if it is possible to mount that backup as the data directory of another standalone mongod instance. I was thinking along the lines of having a dedicated ‘backup’ server that has network connection to the mongodb cluster that needs backing up. This ‘backup’ server provides shared storage to the mongo cluster nodes as supported by PBM . Besides the shared storage the ‘backup’ server also acts as a mongodb node (standalone or single server replicaset).
Is it then possible to let the mongod instance on the ‘backup’ server to use the backup data and start the mongo server with the backup data. Preferably without restoring the backup first to a directory on the ‘backup’ server.
If not, is it possible to restore the backup data to the mongod instance available on the backupserver (could have the same replicaset name). Essentially sending the data to a non-related mongod instance where mongodump could be used to dump selectively.
I hope someone has a useful answer!