So I managed to do something boneheaded today

[Constantin]

(by no means unusual, just happens to be FreeNAS related today)

Got a SSD from ebay, wanted to run badblocks, and ran it on a pool disk instead of the disk I meant to do it to.

FWIW, I'm running bad blocks on the right drive now (the designation changed between reboots from da8 to da5) but I wonder how to fix the damage to the drive accidentally experienced. Now the pool disk (da8) is showing up as DEGRADED (which is a pretty good description of what bad blocks does to a disk) and the question is: how to recover from here? Do I offline the drive and replace it with a different one? Or does ZFS self-repair?

I have several drives I could replace it with, if necessary, which would be offline, swap drives, and then enable, right? However, the GUI reports it cannot online drives for encrypted pools, which mine is. So now what?

 

[Chris Moore]

Do I offline the drive and replace it with a different one? Or does ZFS self-repair?

You can offline the drive and replace it with itself. There is no physical damage, but it will take a resilver to restore the missing data. After you offline it, do a clean to remove the existing partition, then use it as the replacement disk.

 

[Chris Moore]

However, the GUI reports it cannot online drives for encrypted pools, which mine is. So now what?

Sorry, I don't know the answer for that.

 

[Constantin]

I went ahead and offlined the disk and replaced it with one of my cold, qualified spares. It's resilvering now. I did have to enter the passphrase for the pool twice, which was kind of weird, but then I get the following error:

Replacing Disk

Error: [ENOENT] options.disk: Disk not found.

Am now forcing the replacement.

 

[Constantin]

I got several more error messages but now the pool is reporting that it's resilvering and is 1% complete. Something odd is going on in the UI messages but I'm inclined to ignore it all since my intent is to upgrade to 12 soon and then I'll likely nuke this pool altogether and start over to realize the benefits of the special VDEVs in plan to implement.

 

[Chris Moore]

I'll likely nuke this pool altogether and start over to realize the benefits of the SDEVs in plan to implement.

What is an SDEV?

 

[Constantin]

Whoops special VDEV. too tired, Apologies. Corrected above.

My first planned foray will be a three-way 1.6TB mirror pool with each drive partitioned to give ~500GB for metadata and another 1.1TB for small files (have to measure first, metadata needs may be a lot lower after my small-file destruction campaign). Then clear L2ARC to allow non-metadata also. Then see if there is a noticeable difference in performance.