Re: [Jfs-discussion] jfs disk impossible to mount: fsck.jfs run indefinitly without solution!

[Dave K. <sh...@li...>]

On Fri, 2006-12-15 at 12:21 +0100, mrossij wrote:
> Hi,
> I installed Suse 10 ESL connected to an IBM-Storage Disk FAStT600 of 1,9
> TB formatted with jfs file system.
> Yestarday the server chrashed and now is not possible to mount the Disk
> I try to execute:
> fsck.jfs - p /dev/sda1
> but after 22 hours, it's still running!!
> I think It's not capable to repair filesystem.
> This is that appears to video:
> myserver/ # fsck.jfs -d -v /dev/sda1
> fsck.jfs version 1.1.11, 05-Jun-2006
> processing started: 12/15/2006 11.14.6
> Using default parameter: -p [xchkdsk.c:3009]
> The current device is:  /dev/sda1 [xchkdsk.c:1520]
> Open(...READ/WRITE EXCLUSIVE...) returned rc = 0 [fsckpfs.c:3233]
> Primary superblock is valid. [fsckmeta.c:1551]
> The type of file system for the device is JFS. [xchkdsk.c:1537]
> Block size in bytes:  4096 [xchkdsk.c:1850]
> Filesystem size in blocks:  464260419 [xchkdsk.c:1857]
> **Phase 0 - Replay Journal Log [xchkdsk.c:1864]
> LOGREDO:  Allocating for ReDoPage:  (d) 4096 bytes [log_work.c:1783]
> LOGREDO:  Allocating for NoDoFile:  (d) 4096 bytes [log_work.c:1807]
> LOGREDO:  Allocating for BMap:  (d) 1386096 bytes [log_map.c:366]
> LOGREDO:  Allocating for IMap:  (d) 18424 bytes [log_map.c:689]
> LOGREDO:  Allocating for IMap:  (d) 2048 bytes [log_map.c:893]
> LOGREDO:  Allocating for IMap:  (d) 2048 bytes [log_map.c:893]
> LOGREDO:  Allocating for IMap:  (d) 2048 bytes [log_map.c:893]
> LOGREDO:  Allocating for IMap:  (d) 2048 bytes [log_map.c:893]
> LOGREDO:  Allocating for IMap:  (d) 2048 bytes [log_map.c:893]
> LOGREDO:  Allocating for IMap:  (d) 2048 bytes [log_map.c:893]
> LOGREDO:  Log record for Sync Point at:    0x01c01d44 [logredo.c:679]
> LOGREDO:  Log record for Sync Point at:    0x01b81c7c [logredo.c:679]
> LOGREDO:  Allocating for IMap:  (d) 2048 bytes [log_map.c:893]
> ....|.....

This is odd.  Phase 0 should not take very long at all.
> 
> Any ideas!

Did you check the syslog for errors?  I wouldn't expect I/O errors to
result in a hang, but it's worth checking.

If you need to get back online as soon as possible, you can run
"fsck.jfs --omit_journal_replay /dev/sda1".  This will do a complete
check of the file system, so it may take a while, but should at least
make progress.  If you don't have a backup, you may want to mount the
file system read-only and back up anything important first.

Let me know if you have the time to help debug the problem before fixing
it.  I'm not sure that getting a stack trace from gdb will be useful
with the shipped binary.  Building fsck.jfs from source will result in a
binary with symbolic information that may tell us where fsck is hanging
up.  Running fsck.jfs under strace may help.

> Thank.

Good luck,
Shaggy
-- 
David Kleikamp
IBM Linux Technology Center