You are not logged in.

#1 2011-07-20 20:39:59

javagamer
Member
Registered: 2011-03-06
Posts: 5

Random Switching to RO mode on SSD

Today I ran into a problem removing  a package.  I received the error "error: cannot remove file '/usr/': Read-only file system".  After asking what it might be about on the IRC I received the obvious reply "is it mounted ro?".  Maybe, it was because I'd never remounted it as RO, but the thought hadn't crossed my mind.  Sure enough the the drive which holds my root filesystem was mounted as RO for some reason.  Trying to investigate further I noticed something else, many programs (I believe they were all in /sbin/) wouldn't run instead returning "Input/Output Error".  I hoped a restart might just magically fix things so that's what I tried.  Instead, the BIOS didn't even see the drive and it got hung up on "Verifying DMI pool data".  I booted from a USB drive I had and decided to take a look.  I was able to mount both partitions on the hard drive and everything appeared to be there so I tried a restart one more time.

For some reason my hard drive had decided it was willing to cooperate again.  I didn't record how long it worked for, but it was probably an hour or so.  Then, the same thing happened.  Restarting was all it took to fix it this time.  Seeing that the problem didn't appear to be a random, one-time thing I decided to investigate a bit further.  Not really knowing where to start I ran some SMART tests, or tried to.  Smartctl -i reported the drive passed, but it also mentioned two errors which I don't recall.  One of them had to do with a SMART checksum which failed.  I wasn't able to actually run any smart tests as both the short and long tests wouldn't finish and whenever I aborted one (by telling it to run another) the self-test log would contain a new entry about a test apparently at 90%.  Looking in the logs wasn't any more informative to me.  I found the time the first occurrence took place, but nothing before it seemed suspicious.

So, now I'm stuck.  A little information about the drive, it's a OCZ 30GB Vertex SSD which I use for /boot and my root filesystem.  I purchased it just over half a year ago.  Most of my data is on other drives /var, /home, /usr/local, ...  an until now I haven't had an issue with it.  Well, almost no issues with it, occasionally when I would start up I would get repeated messages with something like "link too slow", but rebooting once or twice always fixed those and since I normally have the computer on 24/7 it wasn't much of an issue.  Can someone give me some guidance?  I'm guessing the problem is related to the SSD, but no idea what it might be as I haven't (as far as I know) put any sort of heavy use on it.  It'd almost be nice if the problem was related to something else like the motherboard though because I could use an excuse to upgrade it.

I could use any help you could give.  While the computer is powered off now, I suspect I can get any data that might be helpful from it (log snippets, configuration stuff etc...)
Thanks in advance!

Offline

#2 2011-07-20 20:56:21

JGC
Developer
Registered: 2003-12-03
Posts: 1,664

Re: Random Switching to RO mode on SSD

Probably an issue with your SSD. OCZ isn't the most reliable SSD company these days. My desktop has the same SSD and I haven't been able to boot it since the last corruption. You might try to find out if there's a more recent firmware available.

Offline

#3 2011-07-20 21:34:05

javagamer
Member
Registered: 2011-03-06
Posts: 5

Re: Random Switching to RO mode on SSD

I checked the reviews before buying and it seemed to be good enough.  Also installed the latest firmware when I bought it, but I'll check again now.

Update: No new firmware, good idea though.

Last edited by javagamer (2011-07-20 21:35:33)

Offline

#4 2011-07-21 12:41:38

javagamer
Member
Registered: 2011-03-06
Posts: 5

Re: Random Switching to RO mode on SSD

New information.
I'm currently using the computer with the SSD in it, but I'm fairly certain it will fail again.  Looking through the logs when it started up I found something that could be useful.

Jul 21 02:28:19 localhost kernel: [    3.398138] EXT4-fs (sdc2): INFO: recovery required on readonly filesystem
Jul 21 02:28:19 localhost kernel: [    3.398199] EXT4-fs (sdc2): write access will be enabled during recovery
Jul 21 02:28:19 localhost kernel: [    4.318263] EXT4-fs (sdc2): recovery completej

and from earlier

Jul 20 07:53:18 localhost kernel: [    3.437959] EXT4-fs (sdc2): INFO: recovery required on readonly filesystem
Jul 20 07:53:18 localhost kernel: [    3.438029] EXT4-fs (sdc2): write access will be enabled during recovery
Jul 20 07:53:18 localhost kernel: [    3.647284] EXT4-fs (sdc2): orphan cleanup on readonly fs
Jul 20 07:53:18 localhost kernel: [    3.647354] EXT4-fs (sdc2): ext4_orphan_cleanup: deleting unreferenced inode 1836771
Jul 20 07:53:18 localhost kernel: [    3.647408] EXT4-fs (sdc2): ext4_orphan_cleanup: deleting unreferenced inode 1836772
Jul 20 07:53:18 localhost kernel: [    3.647431] EXT4-fs (sdc2): ext4_orphan_cleanup: deleting unreferenced inode 680913
Jul 20 07:53:18 localhost kernel: [    3.647536] EXT4-fs (sdc2): 3 orphan inodes deleted
Jul 20 07:53:18 localhost kernel: [    3.647593] EXT4-fs (sdc2): recovery complete
Jul 20 07:53:18 localhost kernel: [    3.752842] EXT4-fs (sdc2): mounted filesystem with ordered data mode. Opts: (null)

Not sure if this helps figure out what's actually happening, but it only occurs for the SSD as far as my logs can tell me.

Offline

#5 2011-07-22 08:21:13

stqn
Member
Registered: 2010-03-19
Posts: 1,191
Website

Re: Random Switching to RO mode on SSD

If the SMART self-tests fail/don't finish then I suppose the disk is ready to be returned to its maker... You could still try badblocks, maybe it'll tell you some sectors are definitely dead... Or help reallocate them. I hope you've made a backup already though smile.

Offline

#6 2011-07-23 00:26:38

javagamer
Member
Registered: 2011-03-06
Posts: 5

Re: Random Switching to RO mode on SSD

At the moment it's working again and has been for quite a while.  However, next time it fails I'll run badblocks over it.  Even if it doesn't fail for a while I'm still aiming to get it RMA'd since I don't want it to fail later when I really need it for something.  Luckily there's nothing very important on it, but it would still be a pain to replace if I lost the data.  Before I RMA it I might clone it over to a partition on my data drives and use that to get everything back on my new SSD.

Offline

Board footer

Powered by FluxBB