Stefan Mohn wrote:
Hello,

I'm sorry to disturb you again. My problem still exists with glusterfs2.0.0.
I've tested it with dbench and after one minute glusterfs crashes.

I've used the same setting as described below, but now I've used
glusterfs2.0.0 release instead.
To test the setting i used

dbench -D . 5

the last lines of the serverlog were:
pending frames:
frame : type(1) op(UNLINK)
frame : type(1) op(UNLINK)

patchset: 7b2e459db65edd302aa12476bc73b3b7a17b1410
signal received: 11
configuration details:db.h 1
dlfcn 1
libpthread 1
spinlock 1
extattr.h 1
st_atimespec.tv_nsec 1
package-string: glusterfs 2.0.0

There has to be a  stack-trace following the lines above in the log.
Please paste the stack trace also. The full log below doest not have
it either.

-Shehjar


Here are the Server-Logfiles:
http://www.physik.tu-berlin.de/~frendor/glusterfs/glusterfsd2.0.0.log
(full version)
http://www.physik.tu-berlin.de/~frendor/glusterfs/glusterfsd2.0.0_cut.log
(short version)

Can maybe someone helping me with this issue?

Thanks,
Stefan


Stefan Mohn schrieb:
Hello

I want to use GlusterFS as a nfs substitute for a FreeBSD Server and
Linux clients.
The FreeBSD-server should export several directories. One of these
directories contains the home-directory for the client-mashines.

The problem is that my GlusterFS setup isn't very stable and crashes
quite often. It is less stable, when more than one volume is mounted,
but even when only one volume is mountet it isn't stable at all.

I've tried some different configurations, but now i'm running out if
ideas. I hope you can help me.

my server:
FreeBSD laverne.physik-pool.tu-berlin.de 7.1-RELEASE amd64

I've testet some glusterfs-releases, beginning with Glusterfs 1.4x. I'm
now using glusterfs2.0.0rc4 (also testet with rc2, it's still the same)

How Glusterfs was build:
support packages: bison-2.3_4,1, fusefs-kmod-0.3.9.p1.20080208_4,
fusefs-libs-2.7.3

i've used glusterfs-2.0.0rc4:
export LDFLAGS="-L/usr/local/lib -liconv"
export CFLAGS="-O0 -g -DDEBUG -I/usr/local/include/"
./configure
make install clean
server-vol-file:
http://www.physik.tu-berlin.de/~frendor/glusterfs/new_glusterfs_server.vol
volume homeklein-dir
 type storage/posix               # POSIX FS translator
 option directory /scratch/export2/home      # Export this directory
end-volume

volume homeklein
 type features/locks
 subvolumes homeklein-dir
end-volume

volume server
 type protocol/server
 option transport-type tcp/server # For TCP/IP transport
# option transport-type tcp # For TCP/IP transport
 subvolumes homeklein
# option auth.addr.lscratch.allow 130.149.58.*  # Allow access to "export" 
volume
 option auth.addr.homeklein.allow 130.149.58.*  # Allow access to "export" 
volume
# option auth.addr.lcompile.allow 130.149.58.*  # Allow access to "export" 
volume
end-volume
client:
Linux 2.6.27-11-generic i686 GNU/Linux (it's a Ubuntu 8.10)

Glusterfs:
packages: libfuse2, fuse-utils
it was build as a package with flex, byacc, fakeroot, dh-make from the
source glusterfs2.0.0rc4

client-vol-file:
http://www.physik.tu-berlin.de/~frendor/glusterfs/new_glusterfs_client_laverne.vol
volume homeklein-dir
 type storage/posix               # POSIX FS translator
 option directory /scratch/export2/home      # Export this directory
end-volume

volume homeklein
 type features/locks
 subvolumes homeklein-dir
end-volume

volume server
 type protocol/server
 option transport-type tcp/server # For TCP/IP transport
# option transport-type tcp # For TCP/IP transport
 subvolumes homeklein
# option auth.addr.lscratch.allow 130.149.58.*  # Allow access to "export" 
volume
 option auth.addr.homeklein.allow 130.149.58.*  # Allow access to "export" 
volume
# option auth.addr.lcompile.allow 130.149.58.*  # Allow access to "export" 
volume
end-volume
some logfiles:
I've got 2 sets of logfiles at
http://www.physik.tu-berlin.de/~frendor/glusterfs

The first logfile documents a crash, without any error notification.
I've made some read&write-tests with a client with bonnie++ and read
some mails at another client.
serverside:
http://www.physik.tu-berlin.de/~frendor/glusterfs/glusterfsd.log (73MB)
http://www.physik.tu-berlin.de/~frendor/glusterfs/glusterfsd_cut.log
(shortend to 1 MB)

clientside:
http://www.physik.tu-berlin.de/~frendor/glusterfs/glusterfs_client.log

the second one was when i shutdown the clients. nothing special was
performed. The logfiles give some hints here:

from second_glusterfsd_cut.vol
2009-04-23 22:27:08 D [inode.c:336:__inode_retire] homeklein/inode:
retiring inode(140733193388032) lru=348/1024 active=56 purge=1
pending frames:
frame : type(1) op(LOOKUP)

patchset: cb602a1d7d41587c24379cb2636961ab91446f86 +
signal received: 11
configuration details:db.h 1
dlfcn 1
libpthread 1
spinlock 1
extattr.h 1
st_atimespec.tv_nsec 1
package-string: glusterfs 2.0.0rc4

server-log:
http://www.physik.tu-berlin.de/~frendor/glusterfs/second_glusterfsd_cut.log
client-log:
http://www.physik.tu-berlin.de/~frendor/glusterfs/second_glusterfs_client_cut.log

Thanks alot for you help
Stefan



------------------------------------------------------------------------

_______________________________________________
Gluster-users mailing list
[email protected]
http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users


_______________________________________________
Gluster-users mailing list
[email protected]
http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users

Reply via email to