Replication was working. Source disk dead. How do I safely replace?

[BuildTheRobots]

I have two FreeNAS servers; A and B.
Recently I added one 8TB HDD to each server (single disk stripe) and then set up replication from Server A to Server B.

This has been working fine for a couple of months, but my disk in server A has just died.
This disk has now been physically removed, though I've not yet even detached the volume on server A (so it's currently in an error state). Read only access via server B is working great.

==> Can someone please recommend how the heck I go about replacing the disk in Server A, migrating the data back from server B and hopefully ending up with nice replication working again at the end of it all from A->B (and not loosing the 7tb of stuff that's now only on B)?

----------------
Server A:
FreeNAS-9.3-STABLE-201511280648
HP MicroServer G7 (AMD Turion(tm) II Neo N54L), 6GB ECC.

Server B:
FreeNAS-9.3-STABLE-201511280648
HP MicroServer G7 (AMD Turion(tm) II Neo N36L), 6GB ECC.

Both servers are on the same vLAN and are connected to each other via gigabit switch.

Both servers have exactly the same Volume name and dataset layout, namely:

archive8/replication/media/{film, tv, etc}
Periodic snapshots and replication are recursive and use archive8/replication/media as the volume/dataset top node.

---------
Greatest of thanks for any advice; I'm obviously more than happy to provide more information if you tell me what :)

 

[Robert Trevellyan]

1. Upgrade both machines to the minimum required RAM, which is 8GB.
2. Turn off the replication task on server A.
3. Follow the directions in the user guide for replacing the failed disk in server A.
4. Use either the command-line or a replication task in the GUI to replicate the most recent snapshot from server B back to server A.
5. Not sure the most reliable way to get the A->B replication going again, either reenable the original task, or create a new one.

 

[depasseg]

I've got a post that is similar to this. It's how to migrate to a larger pool, but the mechanics are almost identical.

Edit: Here's the post I was thinking about. https://forums.freenas.org/index.ph...te-data-from-one-pool-to-a-bigger-pool.40519/

 

[BuildTheRobots]

@depasseg: can't obviously see your post, but would happily read it :)

@Robert Trevellyan: As it's a single stripe the freenas handbook seems to indicate I need to destroy it and start again. Would you recommend anything different?

I'm going to give this a go in a pair of VMs so will report back. Thank you both for your replies =)

 

[Robert Trevellyan]

As it's a single stripe the freenas handbook seems to indicate I need to destroy it and start again. Would you recommend anything different?

One of the biggest benefits of ZFS is that it can detect and correct corruption automatically. However, it can only do this when there is redundancy in the pool. Taking a single-disk stripe and backing it up to another single-disk stripe is 100% better than having no backup, but there's nothing you can do about latent corruption. Moreover, if corruption in your primary pool goes undetected, it can potentially be propagated to your backup.

So yes, I recommend you rethink your storage layout.