On Wednesday 12 February 2003 06:09, Juan Antonio Vera wrote:
> Hi all.
>
> I work with kernel 2.4.20 , lvm 1.0 and jfsutils 1.1.1. for 2 weeks
> and I  had  3 or 4 problems with jfs filesystems. The problems
> basically happened
> when I worked with office aplications (StarOffice) or copy files
> between jfs filesystems. The last problem was very serious becasuse
> I've to push reset button to recuperate the system control and I lost
> one filsystem. In this last case, the root filesystem begins to
> report problems about jfs boundaries and
> I press reset button. On reboot, there was one filesystem that jfs
> can't recuperate. This is my fsck.jfs output.

The damage to the volume looks extensive.  Is it possible that the LVM 
volume modified in some harmful way?  /var/log/messages indicates 
problems with the block map.  I see an opportunity to do better damage 
control.  When the first problems were detected, JFS should mark the 
superblock dirty instead of just causing a BUG.  This would force fsck 
to try to fix the volume earlier before more damage could be done.  I 
don't know what caused the earlier problems.  I haven't seen a report 
of the first assert() before.  I've got several things to look at this 
morning, so I may not get to this one immediately.

>
> gava:~# fsck.jfs  -v -a /dev/dades_vg/aplicacions
> fsck.jfs version 1.1.1, 17-Dec-2002
> The current device is:  /dev/dades_vg/aplicacions
> (chklog) FSCK  Open(...READ/WRITE EXCLUSIVE...) returned rc = 0
>
> (chklog) FSCK  Primary superblock is valid.
>
> (chklog) FSCK  The type of file system for the device is JFS.
>
> Block size in bytes:  4096
> File system size in blocks:  90112
> Phase 0 - Replay Journal Log
> (chklog) FSCK:LOGREDO:  Log already redone!
>
> (chklog) FSCK  logredo returned rc = 0
>
> Phase 1 - Check Blocks, Files/Directories, and Directory Entries.
> (chklog) FSCK  Dense file (inode A16) has an unallocated section
> after offset 16.

I haven't seen this before.

> (chklog) FSCK  Primary metadata inode A16 is corrupt.
>
> (chklog) FSCK  Duplicate reference to 2 block(s) beginning at offset
> 42 found in file system object MA16.

This is bad.

> Errors detected in the primary File/Directory Allocation Table.
> Errors detected in the secondary File/Directory Allocation Table.
> CANNOT CONTINUE.
> (chklog) FSCK  processing terminated:  2/12/2003 14.18.57  with
> return code: -10049  exit code: 4.

This looks like it's beyond fixing.  You may be able to recover data 
from it by mounting read-only.

> I'm worry about jfs stability because I'm evaluating journaling
> filesystems and I hear good things about it.

The stability is getting better, and a lot of people are having no 
problems at all, but we're still getting a few reports like yours.

> Any idea about my problems ??.

Not yet, but as I stated above, we may be able to mark the filesystem 
dirty earlier to help contain the damage.

>
> Thanks in advanced.

Thanks for the information.

Shaggy
-- 
David Kleikamp
IBM Linux Technology Center

_______________________________________________
Jfs-discussion mailing list
[EMAIL PROTECTED]
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion

Reply via email to