[OpenAFS-devel] Tracking down openafs-1.2.6-rc1 global-lock-problems

[Herbert Huber]

This is a multi-part message in MIME format.
--------------08F8F764A4994F19DE88D102
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

It seems that our locking problem in openafs-1.2.6-rc1 is memcache
related. We used the following memcache settings:

MEMCACHE="-blocks 128000 -stat 2800 -memcache -daemons 6 -volumes 512"
OPTIONS=$MEMCACHE

Running 4 concurrent instances of bonie++ locks up our machine
reproducibly:
/bonnie++ -d /afs/.../test -s 256 -u <user> -r 128
/bonnie++ -d /afs/.../test -s 256 -u <user> -r 128
/bonnie++ -d /afs/.../test -s 256 -u <user> -r 128
/bonnie++ -d /afs/.../test -s 256 -u <user> -r 128

Please note the the machine doesn't lock up if we change to the MEDIUM
client configuration option:
MEDIUM="-stat 2000 -dcache 800 -daemons 3 -volumes 70"
However the test doesn't complete correctly. All bonnie++ instances die
during the test reading intelligently! Please note that only the user
space program dies. AFS is still up and working.

Performing the same tests on a non-smp linux box (SuSE 7.2, kernel
2.4.18) all bonnie++ tests can be run without any problems.

This last statement tells me that there exists also another smp-related
problem!?


Regards

Herbert




--------------08F8F764A4994F19DE88D102
Content-Type: text/x-vcard; charset=us-ascii;
 name="Herbert.Huber.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Herbert Huber
Content-Disposition: attachment;
 filename="Herbert.Huber.vcf"

begin:vcard 
n:Huber;Herbert
tel;work:089-289-28833
x-mozilla-html:FALSE
url:http:/www.lrz.de/
org:Leibniz-Rechenzentrum;Gruppe Hochleistungssysteme
adr:;;Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschafte;München;;D-80333;Germany
version:2.1
email;internet:herbert.huber@lrz-muenchen.de
title:Dr.
x-mozilla-cpt:;-12480
fn:Herbert Huber
end:vcard

--------------08F8F764A4994F19DE88D102--