I'm looking for some advice before I make things worse...

I run Unraid on a Dell 7060 Micro, which has a single SATA and a single M.2. I don't have a local array - my storage is on a separate Synology with this server just handling containers and VMs, via a cache pool.

Currently I have a single nvme drive providing the cache. I used to have a RAID1 mirror but the second drive was SATA and was getting old so I took it out with the intention of upgrading. Then pricing exploded...

Inevitably, the nvme is now starting to fail, and I need to replace it. My intention is to pickup a larger and better quality nvme more suited for server operations, so I'm just looking at options to handle a migration. The new drive looks like taking a week to arrive, so getting some redundancy back in the meantime would be no bad thing too.

Firstly I tried to recreate the RAID1 mirror, but as the SATA drive is only 480Gb and the nvme as 512Gb it clearly won't let me do that. If I had an array I could backup to that, but I don't.

I think the next best option is probably to clone the nvme to the SATA drive, make that the primary in the pool and then re-add the nvme - as it will be larger than the pool I assume that would be fine?

Then, when the new drive arrives, I assume I'd need to break the pool again to just the SATA, drop in the new nvme, create a second pool at the larger disk's full size (1Tb) and then copy over the data from 1 pool to the other.

Whether any of this is workable or best practice I don't know, so I'm looking for some advice. I have plenty of capacity on the Synology, but I assume that using that isn't sensible, given likely issues with permissions etc when copying between systems?

Any advice gratefully received - I'm a bit anxious that I don't do this right I'm going to take the server down. I do have data backed up, but I still don't really want to be rebuilding it. Thanks in advance for any advice or suggestions!

14 minutes ago, Nuuki said:

make that the primary in the pool and then re-add the nvme - as it will be larger than the pool I assume that would be fine?

Yes, that should work.

15 minutes ago, Nuuki said:

Then, when the new drive arrives, I assume I'd need to break the pool again to just the SATA, drop in the new nvme, create a second pool at the larger disk's full size (1Tb) and then copy over the data from 1 pool to the other.

Do you plan to leave the mirror in place? If not, you can just add the new device to create a mirror, then remove the older one.

Great - thanks Jorge!

So the big question is, what's the right/best way to clone the data across? And just to confirm, as this is the risky bit, are these the correct steps:

• Format SSD

• Create a new cache pool using just the SSD

• Copy data from old to new cache

• Delete original cache pool

• Rename new cache pool to match old (is this needed? Some docker volume mappings use "/mnt/cache")

• Add name to new pool to create RAID1

6 minutes ago, JorgeB said:

Yes, that should work.

Do you plan to leave the mirror in place? If not, you can just add the new device to create a mirror, then remove the older one.

No - the new drive is larger, so I'll run with just that until I upgrade my chassis to something with twin M.2s. Are you saying that I could add the new drive to the pool, remove the old drive and then expand the cache pool to use the full capacity of the new drive? I didn't think that was possible and that a new cache pool was required to change size.

You can use the mover, assuming no root files, just make sure to disable Docker and VM services first, assuming they are using the pool

1 hour ago, Nuuki said:

Are you saying that I could add the new drive to the pool, remove the old drive and then expand the cache pool to use the full capacity of the new drive?

Correct, it will auto expand once the smaller device is removed from the mirror (this applies to btrfs and zfs).