[Gluster-users] Strange performance issues on Debian/Lenny compared to Ubuntu
Olivier Le Cam
Olivier.LeCam at crdp.ac-versailles.fr
Fri Mar 26 11:28:58 UTC 2010
I'm experiencing strange results when comparing the same setup on either
Debian/Lenny and Ubuntu/Hardy.
The server is composed of two replicated nodes (both with Debian/lenny
with GlusterFS 3.0.3, 1GbE).
Config files were generated with the following options:
glusterfs-volgen --name store1 --raid 1 gl1:/export/nunki gl2:/export/nunki
Two clients, one is an Ubuntu/Hardy (LTS, up-to-date) when the other is
a Debian/Lenny. Altough a dd's issued on the Ubuntu client is quite
satisfying, the write performances are disastrous on the Debian client.
ubuntu:# dd if=/gl_nunki/file_test of=/dev/null
524288+0 records in
524288+0 records out
268435456 bytes (268 MB) copied, 3,15924 s, 85,0 MB/s
lenny:# dd if=/dev/zero of=/mnt/file_test count=262144 bs=1024
262144+0 enregistrements lus
262144+0 enregistrements Ã©crits
268435456 bytes (268 MB) copied, 82,408 s, 3,3 MB/s
Notes that might be of interest:
- I have tried to mount the storage directly on both servers and I face
the same poor performances
- I have tried with v3.0.0 at the client side, same result
- I have tried on three different Lenny clients (with different
hardwares), always same results
- With all performance translators disabled, the result is even worse
(less than 1 MB/s)
Actually, I suspect a kernel option in Debian/Lenny to be the cause of
the problems, but TBH I don't know where to focus my attention.
Any help would be greatly appreciated!
Thanks in anticipation,
More information about the Gluster-users