Expand ZFS size

[CLSegraves]

Hello,

I have been searching and can't seem to make my ZFS expand. I originally has a ZFS RAID1 setup with 2 x 2TB drives. I replaced both drives (one at a time) with 4TB drives, yet the ZFS is still the same size. Is there some trick I'm missing?

System info is:
Build FreeNAS-9.3-STABLE-201509160044
Platform Intel(R) Core(TM) i5-2400S CPU @ 2.50GHz
Memory 16266MB


Thanks,
Chris


edit/solved:
I had an older RAID card that limited the format size on my drives to 2.2TB (the remaining space was unallocated). The solution was to:
- remove the RAID card
- connect the drives directly to the motherboard
- one-by-one wipe the drives with DBAN and reinstall, allowing FreeNAS to reformat and resilver the drives to full size

 

[cyberjock]

Did you follow the manual for expanding the volume?

 

[CLSegraves]

Yes. Followed the manual exactly and it refuses to expand (keeps showing 2.ooT total, 1.63T used). Oddly enough, it thinks each disk is 2.2T in size (they are 4TB drives).

 

[Jailer]

Need hardware specs, might be a controller issue.

 

[CLSegraves]

crud, I didn't even think about it being the controller.

I'm using a scsi raid controller as a SATA controller card (not using the raid functions of the card). I'm using the card because it gives me 8 extra SATA ports (MB only has 4), but the card is at least 10 years old, so it might not be able to see more than about 2T.

I'm running a scrub right now, but when it finishes I'll pull the controller to get the data.

I'll move the disks on this volume over to the MB and see if that changes anything.

 

[Robert Trevellyan]

the card is at least 10 years old, so it might not be able to see more than about 2T

That will be the issue. The 2.2TB drive size is a classic symptom.

 

[jgreco]

You may yet be able to rescue this by substituting in a real HBA or standard SATA ports. The problem, however, is that many RAID cards will call it "JBOD" but will still be putting their own spam disklabel on the front of the disk, which means you're dependent on the RAID controller, and worse, in this case, a broken RAID controller.

Consider making a safe copy of your pool back onto one of those 2TB drives, then ditching the RAID controller for standard SATA ports.

 

[CLSegraves]

So the controller (a Super AOC-USAS-L8i) was indeed the issue with not seeing the physical drives' size. I plugged the drives associated with this pool directly into the MB and now the 4TB drives actually show as 4TB.

However, I'm still not having any luck getting the pool to expand. Do I need to pull the drives (one at a time) and then reinstall to force a resilver?

 

[jgreco]

Yeah, um, let me think about that.

The problem is that you already added the drives and therefore they're partitioned at 2.2TB.

You can PROBABLY cheat.

1) If you don't trust me (and I don't trust me, so you shouldn't either) - make a safe copy of your pool back onto one of those 2TB drives.

2) Make sure that the "autoexpand" property is enabled on the pool. CLI: "zpool get autoexpand"

3) Detach one of the 4TB drives. Wipe it clean. I believe the option from the disk management menu in FreeNAS would be sufficient. If not, DBAN in zeroing mode. Reboot NAS (just for safety).

4) Use the volume manager to add that 4TB drive back into the mirror. This will label the drive with a 4TB label and cause a resilver. WAIT FOR IT TO COMPLETE.

5) Make sure it looks OK.

6) Detach the other 4TB drive. At this moment your pool will autoexpand to 4TB. Now finish following the same process to wipe and reattach this drive.

7) Tell me how much of an idiot I am when this doesn't work.

 

[Jailer]

Number one being of the utmost importance in this case. ^^^^^^

 

[jgreco]

Number one being of the utmost importance in this case. ^^^^^^

That's why I followed it up with 7) ;) ;)

 

[CLSegraves]

Trying to replace one the disks (pulled, wiped, reinstalled) and I'm getting the following error:

Error: Disk replacement failed: "invalid vdev specification, use '-f' to override the following errors:, /dev/gptid/b03d80c4-6a29-11e5-98fa-4c72b9e6576b is part of active pool 'Disk02', "

 

[jgreco]

That sounds like it isn't fully wiped.

 

[CLSegraves]

Agreed, but I pulled the disk, stuck it into my desktop, wiped and formatted it to NTSF, and stuck it back into the NAS. So it SHOULD be wiped.

How do I force the override (I would assume that will wipe the drive and then reformat it?)?

 

[gpsguy]

Wiping the drive correctly, should yield a disk that looks like it's never been used.

Try it again using DBAN (dban.org) [as jgreco suggested] this time, with the quick option. It will take awhile, but it's thorough. Don't format it afterwards - use it as is.

 

[CLSegraves]

"it will take awhile" doesn't even begin to describe it. DBAN has been running for over 2 hours and it says 31 hours to go...

 

[jgreco]

Are you running it in zero-only mode?

 

[CLSegraves]

dohh... I had it doing the 3 round wipe (didn't realize there was a single pass wipe). Now it will only take 12 hours... LOL

 

[Robert Trevellyan]

Now it will only take 12 hours...

In this case you can kill it after the 1st pass, which is the write. The 2nd pass is a read/verify.

 

[jgreco]

In this case you can kill it after the 1st pass, which is the write. The 2nd pass is a read/verify.

He wrote that 18 hours ago so presumably it was done anyways ;-)