Replication / Snapshotting retention question

[fullspeed]

Hey Guys,

So it took forever but I got three giant datasets all sync'd and up to date but there has been recently a lot of (temporary) load and replication is a few days behind. It will catch up with this load is off but I need to free up a bit of space in the mean time.

I noticed snapshots aren't being autodeleted right now, they are two days behind (with replication) so I assume they are related. Does the python script that deletes snapshots not run until everything is replicated and "up to date" ?

Can I delete all the snapshots manually to free up space that have already been replicated? for instance it just finished replicating say "02/14/2016 @ 4pm" for a dataset, Can I delete that snapshot and anything earlier than that without repercussions? I am fairly certain I can I just wanted to make sure.

I'm on FreeNAS-9.3-STABLE-201512121950 both push and pull side.

Thanks!

 

[dlavigne]

According to https://bugs.freenas.org/issues/13675 (which you can't see as it's private due to the user's uploaded debug):

The replication system (at this time, at least) does not remove snapshots on the remote system when they're gone from the local system. You will have to, I believe, manually clean up the snapshots on the remote system, just as you did for the local system. (Alternately, destroy the dataset on the remote side, and recreate it, and then let the replication start from scratch.)

 

[fullspeed]

According to https://bugs.freenas.org/issues/13675 (which you can't see as it's private due to the user's uploaded debug):

The replication system (at this time, at least) does not remove snapshots on the remote system when they're gone from the local system. You will have to, I believe, manually clean up the snapshots on the remote system, just as you did for the local system. (Alternately, destroy the dataset on the remote side, and recreate it, and then let the replication start from scratch.)

I was more focused on the push side as that's the one which was backing up, I was right in the sense that once replication was "up to date" in the GUI suddenly all the backlogged snapshots got deleted after being days behind. The question still remains though, Can I delete the backlog myself or is replication reliant on them? I'm assuming anything before the snapshot that is being sent over is deletable but I'm not 100% sure.

In regards to the Pull side/Destination it actually does delete snapshots. Currently there are 96 and I've never touched it, There have probably been thousands of snapshots sent over by now.

 

[dlavigne]

AFAIK it is safe to manually delete the backlog.