Filesystem suddenly corrupted!

Gullible Jones · 2006-04-08 15:50:47

When fscked today, the JFS filesystem on /dev/hda1 turned up errors - rather strange, since I haven't been hitting the interrupt button lately. But I booted up as usual, did 'pacman -Syu', allowed pacman to upgrade my kernel, and rebooted once that was complete... and instead of loading GRUB and booting, it just sat there for a minute, displaying the "loading GRUB..." message.

So booted via the install floppies I always keep handy, loaded up the necessary modules, and tried to mount /dev/hda1, so that I might chroot into it and reinstall GRUB... And instead, I got this:

ERROR: (device hda1): XT_GETPAGE: xtree page corrupt
ERROR: (device hda1): XT_GETPAGE: xtree page corrupt

Yes, I did get it twice. And of course, the mount failed.

So, I ran jfs_fsck again... And this is what it gave me:

Errors detected in Primary File/Directory Allocation Table.
......
File system object FF65726 linked as: /var/run/random-seed
File claims cross linked block(s).
cannot repair FF65726. Will release.

Needless to say, subsequent runnings of jfs_fsck continued to detect this error and failed to fix it. I wiped /dev/hda1 clean and made a new filesystem on it with jfs_mkfs, and that of course got rid of any errors, but before I reinstall I have to ask -

Does this series of events indicate that my primary hard drive - barely more than a year old - is already dying on me? :shock: If not that, what does it mean? Filesystems don't just get corrupt overnight, all by themselves... :shock:

patroclo7 · 2006-04-08 16:15:53

I had a couple of very similar unrecoverable filesystem corruptions with JFS a couple of years ago. I do not remember the details however. Nothing similar happened to me with other linux filesystems.

However, other people say similar things about XFS, which I use flawlessly on a regular basis, so YMMV.

Gullible Jones · 2006-04-08 16:41:57

A virus you say? Huh, I hope I haven't picked up anything. I'm firewalled damn well, haven't been to any hacked or suspicious sites, and driveby downloads aren't so easy to pull off on *nix users... Still, it's a possibility.

(You know, I wish some sort of passive defense mechanism against viruses was available... The way that AVs do things, scanning your disk and RAM continuously, seems like an often-ineffective waste of resources.)

Now I suppose I could see this being a JFS problem... Never had any trouble with JFS before, though, and AFAIK it's pretty stable as of right now.

(XFS I wouldn't use, because its performance with pacman is absolutely hideous - using XFS, I've often had to wait for 5 minutes while pacman was updating its database, searching for a package, or whatever compared to less than 15 seconds with other filesystems... And this even after having the pacman database in the HDD's cache. Sometimes I get the feeling that XFS on Linux has some really serious problems with it. :shock: But I digress.)

But back to the state of /dev/hda... Does anyone here know the average lifespan for one of these? My primary drive is a WD400BB-00JHA0, so I'm hoping that most WD400BBs last for more than a couple years.

(Just FWIW, I've seen older hard drives last 10 years without any problems. I hope they make them to the same standards these days.)

T-Dawg · 2006-04-08 16:57:00

try running a hard drive diagonistic test on it. Most of the time your HD manufacturers will provide this free utility online which you save onto a bootable disk. I know my western digital has a nice one that appears to work for all drives...There's also the altimate boot cd which is free as well.

good luck

iBertus · 2006-04-08 17:07:02

As for HDD life, it really depends. The general rule of thumb is that it's going to happen sooner rather than later. I had a similar drive die about a year ago and it happened in much the way you described above. I'm not saying that your drive is bad, of course you're going to check it before giving up hope.

Gullible Jones · 2006-04-08 20:07:17

Well, if it is bad, that certainly explains the progressively worse performance I've been getting with it. Dang it, I thought WD was supposed to be good stuff...

(IMHO, this sort of thing is why Maxell's new optical hard drives ought to use cubes and mirrors instead of discs and motors.)

raskolnikov · 2006-04-08 20:50:24

You may want to check the smart status of your disk with smartctl (package smartmontools). It contains a daemon (smartd) which can monitor your drives and mail you if there is a disaster in progress, of course it doesn't prevent anything : just tell you it's time to save your data and search for a new drive

Lone_Wolf · 2006-04-08 23:17:59

It would also help to check if SMART is enabled in your bios.

Many bios manufacturers have it default set as disabled.

Gullible Jones · 2006-04-08 23:28:04

Good idea... BTW, I forgot to mention - the Western Digital drive came from my previous computer, a Powerspec 1420 which got fried by lightning. The mobo and CPU were fubar, and the RAM got damaged pretty badly, but the HDD seemed intact, so I continue to use it. I just wonder, could the surge have contributed to what I'm seeing now?

clarence · 2006-04-09 05:33:17

A lightning strike would more than likely have *some* effect i would imagine. Given that situation I would sell that drive really cheaply rather than risk data loss. Damage on mobos, ram cpus etc is really easy to spot but hard drives are a bit more mysterious.

I see a lot of hard drives, and at the moment I would recommend Seagate over WD. Though it used to be the other way around and will probably change again in the future.

patroclo7 · 2006-04-09 08:18:00

( About xfs digression above: in my personal experience (thus, no benchmark), xfs is fast and not-so-CPU-intensive for everything except pacman. So, I have a separate partition for /var with reiserfs (and another one with /boot, since it is not easy to make grub work on xfs). )

lanrat · 2006-04-09 12:02:00

BTW, I forgot to mention - the Western Digital drive came from my previous computer, a Powerspec 1420 which got fried by lightning.

ROTFL
Oh, how could lightning had something to do with this ? :-D

Ok. Seriously, I had at least one quite new WD disk which started to generate irritating noise. No s.m.a.r.t. errors, no nothing in the logs but it's obviously not right. It seem to still work when I'm using it as an external hd through usb2ide but I wouldn't leave it in the server or desktop as a primary disk. I guess things just brake from time to time and you have to get used to it. Anyway, you should use one of the rescue live cds with tools for hard disk testing.

Gullible Jones · 2006-04-09 15:12:56

Eh, ditched it... I'm now using the Samsung drive, which is not giving me any grief (and not producing any audible noise either). At some point I'll test the WD, but right now I've got to reinstall my system, etc.

Bysshe · 2006-04-10 04:53:16

Simply put, I suspect bad pedigree in that puppy. Sparing the details, that's the budget model line of Western Digital. The reputation certainly isn't one of the best, and if Western Digital didn't know it, then they would have given it five years warranty instead of one. :twisted:

Gullible Jones · 2006-04-13 20:32:51

Guess what... You are dead right. Check it out on NewEgg - one good review, and two really bad ones. Talk about a fscking budget product.

(They don't carry 40 GB Samsung HDDs of the type I've got, but the 80GB model is there and got all good reviews, so I suppose I've got at least a few more years with this sucker. Still, backups are the way to go...)

Bysshe · 2006-04-14 00:45:59

Don't put too much stock in Newegg reviews. I do read them, though, but very objectively. There's hardware-centric sites on the net that do more justice.

I didn't realise Western Digital has a newer 3-year warranty now on those. Since July of 2005? The same model standard warranty was previously one year for some time.

The earliest review on newegg's site for that model hails to 4/24/2001. You win some you lose some I guess.

Arch Linux

#1 2006-04-08 15:50:47

Filesystem suddenly corrupted!

#2 2006-04-08 16:15:53

Re: Filesystem suddenly corrupted!

#3 2006-04-08 16:41:57

Re: Filesystem suddenly corrupted!

#4 2006-04-08 16:57:00

Re: Filesystem suddenly corrupted!

#5 2006-04-08 17:07:02

Re: Filesystem suddenly corrupted!

#6 2006-04-08 20:07:17

Re: Filesystem suddenly corrupted!

#7 2006-04-08 20:50:24

Re: Filesystem suddenly corrupted!

#8 2006-04-08 23:17:59

Re: Filesystem suddenly corrupted!

#9 2006-04-08 23:28:04

Re: Filesystem suddenly corrupted!

#10 2006-04-09 05:33:17

Re: Filesystem suddenly corrupted!

#11 2006-04-09 08:18:00

Re: Filesystem suddenly corrupted!

#12 2006-04-09 12:02:00

Re: Filesystem suddenly corrupted!

#13 2006-04-09 15:12:56

Re: Filesystem suddenly corrupted!

#14 2006-04-10 04:53:16

Re: Filesystem suddenly corrupted!

#15 2006-04-13 20:32:51

Re: Filesystem suddenly corrupted!

#16 2006-04-14 00:45:59

Re: Filesystem suddenly corrupted!

Board footer