Can I trust this pool? HDD serial-less after testing.

[StephenFry]

I am currently testing FreeNAS with a small number of drives to see what it, together with ZFS, can do for me. So far, so good.
Today however, I performed (what I thought was) a simple 'disaster scenario', but things didn't go as I expected. It would be great if someone could shed some light on this, I'm sure I'm misunderstanding something.

Setup: 3 x SATA 2TB drive in RAIDZ.
Disaster: a freak accident wipes one of the drives clean.

I restart the NAS, and FreeNAS tells me the pool is degraded. Of course. Data is all still available, good.

Then I performed a scrub. With only 50 or so GB of testdata, this went very quickly and suddenly the red light turns green and tells me the pool is healthy.

But from my (GUI) point-of-view, this is what it looks like:

http://i.imgur.com/sI6f3.png

As you can see, ada0 is the drive I erased. However, the drives are now called *p2 and the ada0 drive doesn't show its serialnr.

edit: I now also see that if I click 'Edit' next to the serial-less drive, the popup screen tells me that the drive is ada2,greyed-out. Now I'm even more confused, since clearly the 'real' ada2 is a drive I didn't monkey around with and is shown, serial and all, in the pool. Uh?!?

edit2: another thing that I just noticed is that when booting FreeNAS, I used to see THREE messages "ATAIDLE: the device does not support advanced power management" and now I see TWO. This is getting weirder and weirder...

edit3: There is no swap partition (of course, since I destroyed the drive's contents), could this be causing problems? And, how to fix?
[root@freenas] ~# swapctl -l
Device: 1024-blocks Used:
/dev/ada2p1 2097152 0
/dev/ada0p1 2097152 0


Question 1: is my pool indeed healthy? Was a scrub all it took to fix this?
Question 2: I thought this wipe would also be simulating a total death of one drive; but I guess should have used a different replacement drive then, instead of putting the original -if wiped- drive back? Since somewhere, FreeNAS/ZFS might recognize the 'old' drive is back and *really* replace by a new one.
Question 3: what does it mean that the ada0 drive is in 'last place' in the Disk View and how do I get FreeNAS to see the drive's serial properly?

Sorry if these q's are silly, but I think after reading all the docs and countless forums, I hoped to have a basic understanding of ZFS/FreeNAS and this test didn't go as expected at all.

 

[peterh]

Yes you can trust this pool if zpool status is ok.

What is confusing you is that freenas wants to partition drived and have zfs use a partition, this would have
happened if you used the procedure described i freenas documentation ( and well described in a tread at this forum)
What heppened here is that zfs happily accepted a whole frive, that's ok but you want see any label-info. And,
if swap was used you woulf have problems since the new drive has no swap allocated.

There is a good thing with partitioning disks before use: one may make shure that the replacement is
equal in size to the failed drive.

 

[StephenFry]

Would your recommendation be that if a disk failed, I would manually partition its replacement prior to adding it to my NAS for FreeNAS to resilver it?

 

[ProtoSD]

If a disk failed, you should use the "Replace" button in the GUI, not scrub, and then it will do all of that for you. You simulated a replacement instead of a failure. There is one replacement scenario that doesn't work right and hopefully will be fixed in 8.2. It involves replacing a disk from a pool created in an earlier release of FreeNAS 8.0. In 8.0 the swap partition was 1GB, in 8.03 and higher it seems to have increased to 2GB. In that case the automatically created swap partition doesn't leave enough space on the rest of the disk to recreate the ZFS partition and the whole replace process silently fails. If that happens, then you'll need to manually create the swap partition. The documentation was updated to cover this situation.