Truenas Core exchange pool with 4+8tb against a 14tb drive

[maeffjus]

Hey Guys, I have two truenas in use, one for my daily use and the other one receives a snapshot with backups from the productive one every night.

Since the second one has some older drives I had lying around, it has 2x4tb and 2x8tb drives in a RAID1 configuration. Total usable space is 12tb.

Now one of the 4tb-drives has (after quite short time - but was used before!) begun to fail.

I ordered 2x14tb drives which shall replace the other ones completely.

How can I do this replacement?
I do not have enough bays to add all drives at once and just moce the data.
So preferably I’d remove half of it and then “rebuild” the data to the new drive, then later remove the other half and replicate to the second new drive.

Any idea how to do this properly?

Regards!

[Stux]

Assuming you want to end up with 2x14TB and 2x8TB in the one pool, just replace the failing 4TB drive with the 14TB. (Physically, and then in the gui)

And when that finishes, do the same to the other 4TB drive.

[maeffjus]

Nope - I want to be left with 2x14tb only.

Currently the data is spread into 11tb of drives (which are a single pool-volume).
I made a mistake - its 2x3tb and 2x8tb.

They shall be gone completely and be replyced by the 14tb-drives.

[Stux]

Scrub the array.

Swap the faulty drive for a 14TB.

Setup a new pool on the 14TB.

Setup a replication from the old pool to the new pool.

Remove the old pool, and add the 14TB in.

Extend the original 14TB drive with the other 14TB drive

As per here:

Expand Single to Mirror?

I managed to double check. So yeah, on Storage Tab, click manage devices for the pool you want to manage. [Screenshot 2024-07-26 at 2.15.45 PM] Then select the device you want to “extend” with another replica (ie another disk to mirror your stripe), then hit extend, then select the drive you wish to use as the additional replica to form the mirror. [Screenshot 2024-07-26 at 2.09.15 PM] “Add VDEV” is not what you want, as that would add another vdev… which would be another stripe in this …

Rename the new pool of you want.

[maeffjus]

Thanks - will do that!

[etorix]

Alternatively, since you have mirrors:
Replace the failed 4 TB by a 14 TB.
Replace the second 4 TB by a 14 TB.
Then remove the 2x8 TB vdev from the GUI.
Do not rename anything…

[Stux]

The above does have a cost of redirect entries in ram.

[maeffjus]

Yes - but when I do - isn’t there any need to move all data from the 2x8tb to the 2x14tb?

From my understanding, the data currently is spread between the 3/4tb drive AND the 8tb drive, this is then mirrored.

So when I remove the failed - sure thing. Then I do the same with the second - fine too. I obviously got the data from two 4tb onto two 14tb.
But the data on the 8tb will stay, right?

And can I eliminate the redicrects later?

[etorix]

Removing a vdev from the GUI (not physically plugging out the drives!) will trigger data migration, and ZFS will eventually offfline the drives when it’s done.