kernel oops under heavy load

[Michael Hanke]

Hi,

My answer comes a little bit late because I was testing the stability with 
respect to load as well to long-term behavior. It seems to work now. The 
reason might be that I had some trouble with the afs-cache. It is long enough 
(2GB). It is an ext2-system but arla resides on reiserfs. Now it uses 
exclusively ext2, and it works... Looks like I have some stale config files 
from my many attempts.

Thanks a lot

Michael

Am Freitag 04 März 2005 09.31 schrieben Sie:
> Tomas Olsson wrote:
> 
> >>Mar  2 18:36:46 struppi kernel:  [pg0+953183561/1069413376] 
> >>nnpfs_fh_to_dentry+0xd9/0x180 [nnpfs]
> >>
> >>    
> >>
> >I think this might be the same thing that Andrew Walters reported on this
> >list a few weeks back. You wouldn't happen to have your cache dir on ext3,
> >would you?
> >
> >Anyway, the output from objdump -S nnpfs_syscalls.o (at least the
> >nnpfs_fh_to_dentry part) would be interesting.
> >
> >Thanks
> >        /Tomas
> >  
> >
> If it is the same problem, I believe I figured mine out yesterday.  It 
> has been up for about 18 hours, much better than five or six minutes.  
> It was one of three problems:
> 
> 1. Make install was broken (again from using checkinstall, piece of 
> crap).  My cache dir was both on my mounted drive and in my arla 
> directory which is on a reiserfs drive.
> 2. It is now working on ext2, instead of ext3
> 3. I bumped up the size of my mounted cache dir from 100 mb to 1.5 gb
> 
> One of my attemps to fix it was putting the entire arla directory on the 
> mounted cache drive and let it use its own arla/cache.  This made it 
> crash within 3 to 4 seconds.  That gave me the idea that it may be 
> drivespace because I just cut down the 100 mb for cache to 44.  Right 
> now my cache dir is only 22 megs, but I would assume it increases and 
> decreases as I write and read from it.
> 
> I get the feeling that it was the last reason, and arla was crashing 
> when I ran out of space on the drive. However, I really dont know which 
> fix got the job done. Hope that helps.
> 
> Drew
> 
> 

-- 
+---------------------------------------------------------------+
|  Michael Hanke                Royal Institute of Technology   |
|                               NADA                            |
|                               S-10044 Stockholm               |
|                               Sweden                          |
+---------------------------------------------------------------+
|  Visiting address:            Lindstedtsvaegen 3              |
|  Phone:                       + (46) (8) 790 6278             |
|  Fax:                         + (46) (8) 790 0930             |
|  Email:                       hanke at nada.kth.se.NOSPAM        |
|                               na.mhanke at na-net.ornl.gov.NOSPAM|
+---------------------------------------------------------------+