You are not logged in.
Posted in "Workstation User" and "Kernel and HW issues"
Hi,
I'm having a problem with ext3 partition and I'm not sure if it's a bug o what. It has happened 3 times yesterday (Sunday), before I switched off the computer at night.
Situation
* A computer (no laptop) with 2 HD (hda and hdd) and a few partitions in each disk.
* Hdd has 1 ext3 and 2 xfs.
* Hda has 1 swap (hda1), 2 ext3 and 1 xfs. The two ext3 are "/" (hda5) and /home (hda6). The rests of directories /var /tmp and so are part of "/", on hda5.
* It's a coincidence (or not) but last weekend I made a "pacman –Sy; pacman –Su", and two of the packages updated were "filesystem" and "kernel". The pacman repositories I have configured are "current" and "extra".
The problem
After a few hours working the console shows the following messages:
journal_bmap: journal block not found at offset 25312 on hda5
Aborting journal on device hda5.
ext3_abort called.
EXT3-fs error (device hda5): ext3_journal_start_sb: Detected aborted journal
Remounting filesystem read-only
From that point, a lot of syslog-ng errors because of read-only filesystem, and also erros from other programs (xorg, kde, etc).
Obviously nothing works properly. In hda5 is mounted "/", including /var and /tmp and it's read-only.
The offset referred (25312) is different each time it happened.
First idea
Could it be the hw, physical error? It could, but when I've had problems with hw, there were lots of messages like "seek error", "read error sector ##", "seek complete error", etc.
But in this situation there aren't. And the error is always in hda5, never in hda6 (also ext3) not hda7 (xfs). So I doubt it's a hw problem.
Why not hda6? My opinion, hda5, root fs, is heavily used compared to hda6 (/home). The third time I got the error, I didn't log into any session. Only the OS and services.
Could it be a bug, related to ext3, after the last upgrade?
Peibol
Offline
Hello,
in the summer I had a similar problem. Unplugging my CDRom and Burner as well as an additional fan fixed the problem. Does the problem you have still occur with only few hardware plugged in? Did you enable SMART for your hds? Run fsck on them.
Offline
Auto answer
I'm going to explain what I did yesterday (yesterday for me; GMT+1).
I booted the PC with a rescue cd. Before doing anything, I tried:
dd if=/dev/hda5 of=/dev/null
and I got lot of errors.
I mounted hda5 and other partition of a different disk, and did a
cp –aR /mnt/hda5/* /mnt/hdd1
and I didn't get any error… so the wrong thing are over non-used inodes.
After that, unmount hda5 and did a
fsck.ext3 –cc
A few minutes, and it finished ok.
Then after doing again
dd if=/dev/hda5 of=/dev/null
I didn't get any error!!!!
The strangest thing was when I tried to get the badblock list, doing dump2fs, but I didn't get any. No badblocks.
After that, I did again a fsck.ext3 –cc, just to be sure, and I did a copy (dd) of the entire hda5, that time without errors.
Conclusion
I'm sure the disk is wrong, physically damaged, and I should change it as soon as I can, but in the meantime I will use it.
Peibol
Offline