[OpenAFS-devel] oops on AFS mount

[Derek Atkins]

Are you running a CACHE in ext3?  That is not supported.  Make
sure your AFS cache is ext2.

-derek

"John A. Goebel" <jgoebel@SLAC.Stanford.EDU> writes:

> Hello,
> 
> We have a problem that _seems_ to be AFS related, although the oops
> points to an EXT3 function.  The problem happens both on openafs 1.2.3
> and release canidate 2, 1.2.4, but not a local filesystem directory,
> nor an NFS mount point. The crash also happens if the machine that the AFS
> space resides is running Transarc AFS or OpenAFS under Linux.
> 
> The test method is to run bonnie++ on a directory in AFS space. 
> 
> After doing a write that is large (>= 1 gig) in AFS space, when bonnie 
> calls an unlink to the created file, there is an oops, and the
> system wedges. While doing a strace on the process I get this:
> 
> temp11 kernel: Assertion failure in journal_write_metadata_buffer() at
> journal.c
> :406: "buffer_jdirty(jh2bh(jh_in))"
> 
> The cache space is at most 60% used, the disk has plenty of space left, 
> and the network to the server side is fine. There is a simple log file
> attached to this mail.
> 
> There was a similar problem the reported on the LKML, at least at the
> functional level <http://linux.msede.com/lvm_mlist/archive/2000/11/0145.html>.
> 
> Attached is the oops, a messy ksymoops of the oops, an strace
> on the running process, and a snapshot of the filesystem state just
> before the crash.
> 
> Thanks for any help you can give.
> 
> John
> 
> ##############################################
> # John Goebel <jgoebel(at)slac.stanford.edu> #
> # Stanford Linear Accelerator Center         #
> # 2575 Sand Hill Road, Menlo Park, CA 94025  #
> ############################################ #
> 
> 
> 

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord@MIT.EDU                        PGP key available