Option to restore one or more db from snap to cluster. Right now, it involves manual dump and restore
Currently we can only restore full cluster from backup like snap using GUI interface. If we want to restore one or more specific db, it needs manual dump and restore from backup. if we have an option to restore specific db to cluster through GUI interface, it will be very useful.
Thirumalaisamy
shared this idea
Hi all – We are currently developing collection and database-level restores. We expect this feature to be released in CY2025, but will update here once it is launched.
-
Nuria
commented
This is very critical for us. Are there any updates about the roadmap?
-
Pablo
commented
The only way to restore an individual collection is restore the whole backup from a snapshot to a new cluster. Then use mongodump tool to get the collection information and restore it in the production cluster.
Deploy a new feature to restore individual collections from a MongoDB Atlas snapshot will be worthy.
-
Matthew
commented
This is absolutely crucial.
-
Gauranga
commented
This is very very critical for us..
PITR is mainly concerned with granular level ( db level, collection level).
-
Eric
commented
This is also crucial for us.
The option to restore just one collection - or even just a limited amount of data from a collection is crucial. The current option to just restore and replace everything is not feasible for many scenarios.
-
Milosz
commented
It's a key core feature for us, bumping as well.
-
Geoffrey
commented
I vote for this feature again.
This feature is very useful in the context that we need to recover or revalidate the contents of a database at a specific time without having to recover the entire cluster.
Saves time and money.
Hoping that this time will be the right one.
-
Matthew
commented
I'm bumping this, as perhaps after 4 years, someone will check this out.
-
RuthvikReddy
commented
Hi all,
There used to be a queryable backup option back in the day,
Would love to see if it can be brought back!This is a key feature that is required !!!
-
Matthew
commented
Upvoting, because a large health insurance client is specifically requesting this feature.
-
Giuseppe Iamele
commented
agree, feature (specify specific database or collection) it was requested by our customer to optimize cost and time to take/restore backup.
-
Andrei Mihalciuc
commented
Current process of restoring a full cluster snapshot requires a lot of time (which is critical during DRP) and resources. It also very brittle due to a number of manual steps required. Having a possibility to restore specific collections would simplify the process significantly.
-
Lennart Tondeleir
commented
This would be a very useful capability to have.
-
Lennart Tondeleir
commented
This feature would really help smoothen out DRP procedures. Upvoted!
-
Jim
commented
This would be a great feature to have.
-
Jim
commented
This would be a very useful feature for us.
-
Ola
commented
Upvoting the request, received the feedback from Customer that this feature would be very useful
-
Sanket
commented
This is important feature request to have backup at collection level. Everybody does not need snapshot of the entire database which could be of very large size and If I want to restore specific collection then there is a long process of getting rid of all non-required collections.
-
CMI
commented
I second this request. Due to the backup being a full cluster snapshot, including oplog and other internals, its size is much larger than our actual data. Being able to just backup the contents of one or more databases would lead to smaller backups that are likely also faster to restore.
-
Geoffrey
commented
Hello,
The Atlas portal/API offers the ability to restore a complete database instance from snapshot or restore point. However the scope of this restoration covers the entire instance.
It would be a great asset to be able to restore a simple database from snapshots (encrypted or not). Currently, restoring a database (not a cluster) is a heavy workload that requires outputting data to a local workstation.
Thanks.