Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I want a shared filesystem on top of drbd8 on two nodes. The servers run ubuntu 9.10.

I googled a lot, but couldn't find an clear trend what the web community prefers. It's seems that OCFS2 is more used at the moment.

Which filesystem is more reliable, faster? GFS2 or OCFS2? Is the linux community going more towards GFS2 or OCFS2? Which of this two is better supported by ubuntu 9.10?

Are there better (or more common) alternatives?

share|improve this question
Im specially interested in answers from somebody who has used both, and can compare them. – yvess Feb 1 '10 at 17:22
I settled on ocfs2. – yvess Feb 2 '10 at 14:10

We use GFS2/rgmanager/DRBD and KVM for our cloud infrastructure.

I don't know what sort resource manager OCFS2 has; but gfs2 rgmanager is great and supports live migration; which work flawlessly.

I have been meaning to try out GPFS but have not got around to testing this; I was going to setup gpfs nodes in the existing kvm/gfs2 setup as vm guests.

Overall: Very reliable, some performance concerns with GFS2 locking between nodes even with rate limiting disabled.

  • One node plock test

[root@core-01 ~]# ./ping_pong /gfs2-00/test 3

2159 locks/sec

  • On both nodes

[root@core-01 ~]# ./ping_pong /gfs2-00/test 3

1336 locks/sec

[root@core-02 ~]# ./ping_pong /gfs2-00/test 3

1333 locks/sec

  • One node pclock rw test "./ping_pong -rw /gfs2-00/test 3"

[root@core-01 ~]# ./ping_pong -rw /gfs2-00/test 3

2192 locks/sec

  • Two node pclock rw test "./ping_ping -rw /gfs2-00/test 3"

[root@core-01 ~]# ./ping_pong -rw /gfs2-00/test 3

  2 locks/sec

[root@core-02 ~]# ./ping_pong -rw /gfs2-00/test 3

  2 locks/sec

I attempted to document some of our work:

share|improve this answer
if those are your lock numbers, you have some serious config isues: root@c1ws1:/gfs/test# ping_pong test1 10000 113588 locks/sec root@c1ws1:/gfs/test# root@c1ws2:/gfs/test# ping_pong test2 10000 114021 locks/sec root@c1ws2:/gfs/test# root@c1ws1:/gfs/test# ping_pong test 10000 5354 locks/sec root@c1ws1:/gfs/test# root@c1ws2:/gfs/test# ping_pong test 10000 6675 locks/sec root@c1ws2:/gfs/test# Can a moderator please merge this as a comment on testlabs answer, I don't have enough rep to comment directly. – karmawhore Jun 25 '10 at 10:53
@karmawholre Your numbers probably will be the same as mine. Thanks, - Adrian – user82323 May 23 '11 at 8:09

I can't talk for everyone but certainly we're very happy with OCFS2, whether that be on bare-metal OEL or within ESX VMs.

share|improve this answer
Did you also look at GFS2? Where there any arguments which let you not consider GFS2? – yvess Feb 1 '10 at 17:23
it's just we have a global license with Oracle for all their code, so use OEL over RHEL or Centos for instance, it means we have free OEL support all the time - which is a bonus :) – Chopper3 Feb 1 '10 at 17:24

I tested both GFS2 and OCFS2 on a drdb Ubuntu-9.04 storage server pair, with a pair of Ubuntu-9.04 servers mounting the kvm images via NFS and iSCSI.

I gave up configuring GFS2 as the instructions for OCFS2 worked almost out of the box for me. I still don't know why GFS2 did not work as I have set it up in a pair of virtual machines since then without a problem.

We are quite happy with the performance of drbd + OCFS2 + kvm. We did run into problems early on with short drbd timeouts though. If we used qcow2 expand on write images, then we would get intermittent VM crashes under high load. Since converting the images to raw we have not had any problems.

share|improve this answer
Why not use plain drbd instead of raw images? It's not like you're saving any space. – ptman Sep 1 '10 at 19:15

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.