[OpenAFS-devel] AIX system hang

Hans-Werner Paulsen hans@MPA-Garching.MPG.DE
Thu, 10 Apr 2003 09:08:09 +0200


On Thu, Apr 10, 2003 at 12:06:43AM -0400, Derrick J Brashear wrote:
> On Wed, 9 Apr 2003, Hans-Werner Paulsen wrote:
> > again we had system hangs on two machines running AIX-4.3.3 and
> > OpenAFS 1.2.8.
> > The last years we saw these problems only on multiprocessor machines
> > using the memory cache. Therefore I created a RAM-disk and used this
> > as AFS cache partition. But now we had system hangs with this
> > configuration: memory cache is NOT the reason for the system hang.
> 
> In the interest of narrowing it to one thing:
> -does it happen on a uniprocessor system?
> -does it happen if you use a disk cache backed by actual disk?
> -is there any version of AFS (IBM, Open) that does not have this problem?
> 
Dear Derrick and other OpenAFS developers,
- we saw this problem only on multi-processor systems
- we never saw this problem using a "real" disk as cache
- beginning with Transarc AFS (I do not remember the version number) up
  to OpenAFS 1.2.8 we had this problem with each version we tried (about
  15 different versions).

Sometimes the system hang happens when the system is started (when
the first AFS-file is accessed). Sometimes we have system hangs
when the system is running from 1 week to several months.

-- 
Hans-Werner Paulsen		hans@MPA-Garching.MPG.DE
MPI für Astrophysik		Tel 089-30000-2602
Karl-Schwarzschild-Str. 1	Fax 089-30000-2235	
D-85741 Garching