[OpenAFS] afs_syscall hangs

markus.hetzenecker@uibk.ac.at markus.hetzenecker@uibk.ac.at
Tue, 4 Nov 2003 23:28:59 +0100


hi all

first of all, thanxs for the great job developing OpenAFS.
we are hosting 2000 home volumes for linux users and everthing worked fine  
till one month ago, then this problem occasionlly happens on a client:
  a process hangs and cannot be killed, in no way.
if i execute at such a client in my home dir:
$ strace fs  flushvolume
then i the process hangs at the following line
afs_syscall(0x14, 0x80a4960, 0x400c5625, 0xbffff150, 0

or i execute 
$ strace find .
then i hangs at some file of my home dir.

if i go with ssh to another client, performing the same commands, everthing 
works fine.
to get rid of all the hanging processes at the client, i need to restart the 
fileserver with "bos stop|start localhost fs".

the problems started with openafs 1.2.9 and doesnt vanish with 1.2.10
server: linux redhat 9, kernel 2.4.20-18.9smp., openafs 1.2.10
clients: linux redhat 9, kernel 2.4.20-20.9, openafs 1.2.10

its not much i can tell about it, but it`s annoying. nothing appears in 
log-files and i found nothing about it in the mailing lists.
please tell me, what to report the next time it happens.

thanxs a lot, markus.
-- 
fork() off;
GnuPG key: http://www.uibk.ac.at/~csaa5103/public_key.asc