Arch Linux

Berticus

Member

Registered: 2008-06-11

Posts: 731

[SOLVED] fsck.jfs hangs

I have no idea what happened. I was swapping my hard drives around as I was moving into the final stages of my migration. From the moment I hit the power button, something seemed terribly wrong. Getting through the initial BIOS check took forever, and I finally met a message complaining about the 3rd IDE disk's S.M.A.R.T. command failed, plus it sounded like the hard drive was having trouble spinning. I hit F1 to continue anyway, which was a terrible idea.

When the boot process was finally handed to the kernel, I met some new checks I had never seen before. Unfortunately I wasn't sitting right in front of the monitor (tinkering with the other side of the computer), but I remember seeing something along the lines of hardware being swapped and getting the filesystem remapped or something along those lines. It finally got to the "loading spamd" part, and then it just stopped.

I didn't want to hard reboot or anything because I'm running an LVM jfs. Even though nothing had happened, except for that odd filesystem remap thingy, everything should be intact. Since nothing was happening, I hit the little button.

Now it can never seem to get through the fsck.jfs for /dev/mapper/vga1-usr. /dev/mapper/vga1-opt and I believe /dev/mapper/home (encrypted and LVM) passed fsck.jfs without a problem. /dev/mapper/vga1-usr is only 5 GB; I left it alone for half an hour to see if it needed more time. It didn't output anything.

Another odd thing is I reverted my setup to the way it was, and the hard drive in question didn't have any spinning problems. The only difference is it's standing up portrait-wise instead of landscape-wise, if that makes any sense. So I'm quite perplexed as to why changing the orientation of my hard drive would cause it to have such problems.

I booted up into the setup CD to see if maybe the fsck on the hard drive was having problems. I didn't leave it alone for quite as long, but it seemed to have the same problem. Nothing was being outputted to the screen, and it wasn't going anywhere.

Do I just need to leave it alone for a few hours to repair or whatever? What are my other options?

I was thinking about reformatting /dev/mapper/vga1-usr and then re-installing everything or something like that. Maybe in the setup CD I can mount it somewhere and move all the files to another partition, reformat it, and move everything back on. I'd just need some way to skip the fsck.jfs.

Last edited by Berticus (2008-08-13 16:45:36)

moberry

Member

Registered: 2008-08-08

Posts: 16

Re: [SOLVED] fsck.jfs hangs

Drive could be on its way out, and the change in angle is enough to keep it going. Never heard of it happening. But I've seen crazier things.

Berticus

Member

Registered: 2008-06-11

Posts: 731

Re: [SOLVED] fsck.jfs hangs

Well that sucks majorly considering I got it about a month ago

moberry

Member

Registered: 2008-08-08

Posts: 16

Re: [SOLVED] fsck.jfs hangs

Common thinking is if a drive is going to fail, it will fail within 1 year. Past that the faulure rate is very low.

pheon

Member

From: Berlin, Germany

Registered: 2008-05-14

Posts: 91

Re: [SOLVED] fsck.jfs hangs

Common thinking is if a drive is going to fail, it will fail within 1 year. Past that the faulure rate is very low.

This only applies to HDDs allready shipped "damaged" (I wonder how they pass manufacturer's qualititytest).
If there is no special circumstance (e.g. physical force (>350g) applied to the device or your spilled beer over it) the MTF (Mean Time to Failure), most manufacturers state, exceeds one year of "normal" use (I am not talking about 24/7 with high read/write frequencies like in servers) and can be considered as a "good average" - of course it is just a statistical value...

In Berticus' case I'd recommend not to "play" around with the HDD until someone provides a way to access the volume (and copy as much data as possible). Every power-on or access to the HD could make the problem even worse.

---

watching someone else use your computer is like watching a drunk orangutan solve a rubix cube

Berticus

Member

Registered: 2008-06-11

Posts: 731

Re: [SOLVED] fsck.jfs hangs

I doubt it's a hard drive failure. At least a permanent one. It seems everything is fine now (passes the SMART check and it sounds like everything is spinning just fine), it just hangs at the fsck.jfs. So maybe that one filesystem is messed up. I was reading online someone said that it can take a while for a jfs to repair itself and the time varies depending on what's on the disk. I ran a fsck on the other filesystems and they all passed quite quickly. I haven't mounted any of them yet, but I don't think I'll have problems doing so.

My two other hard drives should make up enough space to back up everything on the new hard drive.

Would be nice to have solid state...

moberry

Member

Registered: 2008-08-08

Posts: 16

Re: [SOLVED] fsck.jfs hangs

Try erasing all your partitons, recreating them, and mkfsing again.

Berticus

Member

Registered: 2008-06-11

Posts: 731

Re: [SOLVED] fsck.jfs hangs

ehh... moving around 1 TB isn't exactly the funnest thing to do... I've done it before and am hoping to avoid it.

Anyway, I believe I know how to recreate and get rid of the problem. It seems the issue is coming from my new PSU. I got a Enermax Modu82+. The sata power cables allow up to 3 disks. My setup before had 3 disks on one cable and the new hard drive had a dedicated cable while I was working on it. When I switched things up, I took out the smallest capacity hard drive (250 GB) and put the 1 TB hard drive in its stead. That seems to be the only time I have an issue with the SMART command, and the spinning problem occurs. Last night what I did was I put it back on its own dedicated power cable and everything was working swimmingly except for the damaged filesystem.

Unfortunately it means I have to re-install just /usr. The only way I know how to do this is to take one of the other hard drives and have it setup like a new system, except use the faulty /usr on this hard drive. Does this make sense? Or is there a faster way of re-installing my programs?

Oh, I attempted to mount /usr and it suggested I dmesg|tail, so:

#dmesg|tail
md: raid5 personality registered for level 5
md: raid4 personality registered for level 4
md: raid10 personality registered for level 10
device-mapper: ioctl: 4.11.0-ioctl (2006-10-12) initialised: dm-devel@redhat.com
SGI XFS with ACLs, security attributes, realtime, large block/inode numbers, no debug enabled
SGI XFS Quota Management subsystem
Filesystem "dm-4": Disabling barriers, not supported by the underlying device
XFS mounting filesystem dm-4
Ending clean XFS mount for filesystem: dm-4
JFS: nTxBlock = 8192, nTxLock = 65536

I find it odd that the PSU had no problem supporting 3 disks with 750 GB, 120 GB, and 250 GB, but had problems with 750 GB, 120 GB, and 1001.1 GB.

moberry

Member

Registered: 2008-08-08

Posts: 16

Re: [SOLVED] fsck.jfs hangs

I dont see any mount errors in that output? Is it further than 10 lines up? PSU would be a good solution to that problem

as far as reclaiming a messed up /usr.. If you dont have a copy of it anywhere, re-installing might be faster. If you just cant do that. install a system to a different PC or drive. Install all packages listed in your current pacman -Q. copy the /usr over to the new drive. In theory that should do it. Which is about what you said, except installing all packages listed in your pacman -Q

usr just contains binaries and libraries, so mimicking all packages installed should create it close enough

Berticus

Member

Registered: 2008-06-11

Posts: 731

Re: [SOLVED] fsck.jfs hangs

Yeah, I'm not exactly sure what nTxLock and nTxBlock is, but mount complained of a bad superblock. It wouldn't be 10 lines up further while in the setup CD. I'd have to look in the logs to see what it was complaining about with the system remap thing.

As for the PSU, Enermax is right up there with Antec and Corsair. It's the 625W versions, so the only thing I would be concerned about is if it wasn't outputting enough power altogether. In that case I would either need to update everything (the motherboard I got is very power hungry), upgrade to the 825W version of the Enermax Modu82+, or get something from the Antec Signature series (~$299). But yeah, Enermax is a very good and well known brand. Which is why I find it odd that it would have issues housing 3 SATA drives on one cable when it was doing that before. Oh I suppose I did change a little bit of which cable is (it's a modular PSU), but that shouldn't make a difference either. I mean there are a lot of good solutions, the one I present is just the cheapest.

Well I know what usr contains, I was wondering if pacman is in /usr/bin (which is what I'm assuming). If it was located in either /bin or /sbin, that would be a lot more convenient, so I could just chroot into my setup and work in there. I'd just be able to do something like pacman -Syu and I think everything would be able to work.

Another thing is where does pacman read from with the -Q option? Because then I would have a list of software I used to have. I wouldn't have to rethink everything. I mean convenient I have a short list of software I had, but it's the very basic (urxvt, firefox, xmonad, pidgin, et cetera). My guess would be to look in /var. Of course not having my arch linux setup, I can't really check (plus I'm at work).

And then there were some programs I installed from source (aur was behind). Unfortunately I moved the aur builds from ~/.builds to /usr/local/builds.

---Edit---
I just found http://wiki.archlinux.org/index.php/How … l_Database, and that's really the bee's knee right there since my /var is intact. Now all I need is pacman. I suppose I can mount root and see if it's installed in /bin or /sbin. If not, I should be able to just grab it from the installation CD.

Last edited by Berticus (2008-08-13 16:44:51)

moberry

Member

Registered: 2008-08-08

Posts: 16

Re: [SOLVED] fsck.jfs hangs

whereis pacman says /usr/bin here

you could do it from the live cd.