Imported zpool from previous freenas setup is not visible from the gui

[Senar]

FreeNAS-8.3.1-RELEASE-p2-x64

I lost my previous freenas setup when the usb stick i used as system disk failed (stupid decision but i had no choice at the time and was too lazy to change it).

I had a zpool made of 4 disks.

I don't know if it has any implication but the zpool was scrubbing at the time the system failed.

When i boot from my new freenas setup, the zpool is at first invisible, simple import command doesn't work, as well as an auto import from the gui.

However i'm able to recover my zpool (mounted and readable/writable) if i use this one :
#zpool import -d /dev -f -a

Nevertheless, the zpool is still not visible from the gui and the auto import doesn't work even if i export the zpool. And of course it disappear if i reboot the system.

At least i haven't lost any data but there must be something i can do that would allow me to import the zpool to the gui?

 

[cyberjock]

Can you elaborate on what was going on when the system failed? Did you lose power or did it crash.

 

[Senar]

I've read it several time, i don't think that i've missed any relevant information.

 

[Senar]

Can you elaborate on what was going on when the system failed? Did you lose power or did it crash.

The system just froze. I was playing a video file from it and it stopped. When i tried to restart it, the usb stick wasn't bootable.

I installed freenas on an hard drive and from there you already know what i did.

 

[cyberjock]

Did you look on the monitor before you rebooted it to see if it had a kernel panic message or anything?

 

[Senar]

I didn't have that chance.

 

[Senar]

I've just seen that the disk occupation of my zpool is correctly reported in the repport section of the GUI. It's taunting me :( !

 

[cyberjock]

LOL.

I'd recommend you do a scrub of your zpool.

 

[Senar]

As i said, the pool was scrubbing at the time it crashed and it continued when it went back online. It did find some errors and the smart service is signaling bad sectors, but there is no way to know if this is a consequence of the crash or if it preceded it as before the crash the smart service wasn't working properly.

As the system was completely frozen i had to cut the power and one of my disk may not have liked it very much. I will pull it out of the pool and replace it if i find a solution or simply build a new pool without it if i can't make my actual pool visible from the gui again.

 

[cyberjock]

But didn't you just say that the " disk occupation of my zpool is correctly reported in the repport section of the GUI.". I took that to mean you got the zpool to mount. Did you not?

 

[Senar]

There seems to be a misunderstanding, as i said in my first post, i can mount the pool from the shell with the import command as long as i point it to the /dev and -f -a (i suspect that the force argument is useless in that case but didn't test without it). I'm actually making a backup of the file trough ftp.

But the zpool is not visible from the web interface, in the volume management or the auto import (even after exporting the pool), wich prevent me from managing it.

I had just seen that despite being invisible from the management tool, it is shown in the repport section, wich is kind of frustrating.

 

[cyberjock]

Oh, ok.

If you don't(or can't) mount the zpool from the GUI then FreeNAS' middleware doesn't know to monitor the zpool.

As far as fixing your problem, once you had chosen to "force" the zpool to mount with the -f command all other options are out the window. In terms of data recovery when there's a problem with a zpool, the -f is one of the last things you should do because its undoable and can damage the zpool. Generally, at the situation you are in, other people are backing up their data so they can destroy and recreate the zpool from scratch.

 

[Senar]

Well that's what I'm doing right now, there is a fair amount of data so it may take a while, that's why i was trying to find an other way around... Now at least i know i don't have to hope for an easy solution.

I should have tempted to import without the -f. But i didn't read any clear warning about it anywhere until now, it still seems a bit odd that it would have that kind of impact as the pool seems to work perfectly fine. Is there any kind of documentation about this behavior ?

 

[cyberjock]

Behavior in what way? You mean warnings about the -f? There are in some places, but its really more of a "you shouldn't be doing stuff until you understand what the consequences are". In your case, you probably did a little googling, found someone that used -f and tried it yourself. Then, after you patted yourself on the back for saving your data I threw some egg your way when i said you took a risk. There were other options available before you got to the -f option, and you might have a zpool that would be fully functional now, but its a little late because you did do -f.

The first step when you have a problem is to diagnose the issue. Then worry about how to correct it. You(and many people) panic when they see that their data is missing. Their life flashes before their eyes as they mentally try to figure out if they have copies of some of their data anywhere else. The -f is mostly a 'hail mary' and is a "im desperate dammit.. give me my data if you can" approach. It can go well, or it can go horribly. There were lots of other potentially less risky options you probably could and should have tried, but most people don't want to spend hours(or days) reading up on what to do and to do it properly. Instead they(probably you too) were in a panic and wanted to see your data. Unfortunately, taking the "ZOMG get me my data now" is precisely when you shouldn't be taking risks.

Keep in mind that even after all this is done, you still should figure out why it froze. The last thing you want is to trust hardware that is starting to fail.

 

[Senar]

There is no need to be so melodramatic, of course i was anxious about my data, who wouldn't be ? The system crashed 2 weeks ago, i did try other things and spent a considerable amount of time trying to find a solution, and didn't find a lot of informations that seemed to be relevant. A the time it didn't appear to me that i had a lot of options. It's too late this time but what can i read to have a better understanding of what happened ?

As for the diagnostic i already told most of it : the usb stick that hosted the system is dead and there was no monitor connected to the computer when it crashed, i can't connect one if i don't move it first, so i don't know if there was any error messages. All i can add to that is that one of the disk has now unreadable sectors. I had no warning from smart before the crash, but i think my hardware prevented it from working correctly (atom motherboard, the doc is a bit ambiguous but it seems that only the port connected to the chipset are smart capable).