__realm_sync history taking up all the storage on Atlas cluster

I recently noticed that the history collection under __realm_sync has quickly taken up ALL the storage in my cluster, which seems excessive. The actual data that I’m syncing is extremely minimal (not even 1 MB).

Is there a way to handle realm’s history better so it doesn’t consume so much storage?

See here (history)

And here are the 2 collections for my actual data:

1 Like

@Annie_Sexton Realm Sync stores all the operations that have mutated any data over the course of the lifetime of your Realm Sync app. You can see more details here - https://docs.mongodb.com/realm/sync/protocol/#changeset

We do have a compaction feature that we have rolled out to some users which will trim some operations from the history collection. If you’d like to have it enabled for your Realm Sync app please send me an email with the URL of your Realm App dashboard and I can get it enabled for you - ian.ward@mongodb.com

Yes - this has been a problem for me as well. I’ll send you a note to get the compaction features rolled out on my dev and prod instances.

What are the implications if I just go in and drop all from the collection? I presume this feature allows for some rolling back, which I guess leads to another question of how you can actually roll it back?

If you drop a collection you will break sync and will need to terminate and re-enable sync - https://docs.mongodb.com/realm/reference/terminating-and-reenabling-realm-sync/

I am facing this same issue in Aug 2022 on a Shared Cluster. Has this this compaction feature already rolled out publicly? Or do we still need to reach out to you to enable it on your end?

I am facing the same issue and it seems flexible sync could be a solution because it supports “trimming” but I could not find a migration guide to migrate partition sync to flexible sync.