/sbin/gfs_mkfs -p lock_dlm -t vmcluster1:vmfs1 -j 8 /dev/sdb

/dev/sdb /gfs2 gfs rw,hostdata=jid=0:id=65537:first=1 0 0

Start /sbin/fenced

Then start the following services:
rgmanager
gfs
gfs2
cman
clvmd

/sbin/fenced
/etc/init.d/gfs start
/etc/init.d/gfs2 start
/etc/init.d/rgmanager start
/etc/init.d/cman start
/etc/init.d/clvmd start

cman tends to hang in fenced until the other node starts.

LVM how-to
http://tldp.org/HOWTO/LVM-HOWTO/benefitsoflvmsmall.html


lvcreate -L4G -nlv1 vg1
lvremove to undo.

mount -t gfs /dev/vg1/lv1 /local/lv1

http://sourceware.org/cluster/wiki/DRBD_Cookbook

A.2. Sharing LVM volumes  (from LVM-Howto)

[root@CentOS-5 rc.d]# vgchange -cy /dev/vg1
  Volume group "vg1" successfully changed
Hi Nirmal,


I think what happened here is that maybe you have locking_type =
2 but clvmd isn't running. Therefore,
lvm can't do the clustered locking necessary to "see" the
volume group, and it has to "see" the volume group before
it can change its attributes.

Your choices are to (1) activate the clustered locking

mechanism and leave the volume group clustered, or (2) change
the locking_type = 1 in your /etc/lvm/lvm.conf file, at least
temporarily, so you can

change the clustered bit off.

To get the clustered volume to be active, make sure
locking_type = 2 (for RHEL4 and similar) or locking_type = 3
(for RHEL5 and similar) in /etc/lvm/lvm.conf, then you need
to run the clustered LVM manager: service clvmd start.
The clvmd service can only run properly, however, if the rest
of the cluster infrastructure is running.

If you're sharing the volume group between systems, you
want to keep the clustered bit on.  If not, you can turn it off.

After clvmd is started, or locking_type changed back to 1,
you should be able to do what you want.  For example:

vgscan
vgchange -aln

vgchange -cn rac

Regards,

[root@CentOS-5 rc.d]# man gfs_tool
[root@CentOS-5 rc.d]# gfs_tool list
4175118336 253:2 virtual_cluster:vmfs1.0