On Sat, 2010-03-27 at 13:36 +0200, Cheusov Aleksey wrote:
>         I have no idea what caused the trap.  It happens in stable
>         code that
>         hasn't been changed in ages.  It could be due to a memory
>         corruption bug
>         somewhere else in the kernel, either somewhere else in jfs or
>         elsewhere.
>         Otherwise, it's something really subtle that I haven't seen
>         before.
> 
> Ok, can vserver affect JFS's stability and in general what kernel[s]
> you can
> recommend for use with JFS?

I don't know of any reason that vserver would hurt JFS.  The code itself
has been stable for a number of years now, and not much has changed, so
I don't wouldn't recommend one kernel over another.  "Stable" here has
more to do with the fact that it hasn't changed much, rather than that
the code is rock-solid.  There have been some elusive bugs reported that
nobody has completely figured out.
> 
> This fs contains  69955977 files and hardlinks. Can it change
> something?

Can what change what?
>         
>         It couldn't allocate enough memory.  It seems odd since you
>         said that
>         you ran version 1.1.11 recently.  I don't know what would have
>         changed
>         that would consume any more memory.  Maybe more inodes have
>         been created
>         since the last attempt?
> 
> See above. Almost 70mln of files. And yes, since last 'fsck.jfs -f',
> more inodes and more files
> were created during nightly backup.

> 
>          Is it possible to add some more swap space and
>         try again?
> 
> This machine have almost 5Gb of swap. Is this not enough for fixing
> 1.5Tb disk partition?

If I calculated it right, it looks like jfs_fsck needs 192 bytes per
inode (file or directory).  So 70 million files would require over 10
GB.  How much main memory does the machine have?  I'm assuming a 64-bit
executable, since a 32-bit binary doesn't have the address space for the
job.
> 
>         Have you tried mounting read-only after the reboot?  That's
>         not the best
>         solution, but at least you may be able to recover the data.
> 
> This is actually a backup disk.

So worst case is to blow it away and recreate it.  :-/

Shaggy
> 
-- 
David Kleikamp
IBM Linux Technology Center


------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Jfs-discussion mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jfs-discussion

Reply via email to