GFS: configuration/understanding problem in getting pools up on all nodes

From: Christian Horn (christian_horn_at_gmx.de)
Date: 12/31/04 

Date: 31 Dec 2004 11:49:04 GMT

Mornings,

i read the default redhat-gfs6-documentation
(rh-gfsg(EN)-6.0-HTML-RHI (2004-06-01T12:12-0400)) and older sistina-
documents.
Installed three RHEL 3.0 u4-boxen, downgraded the kernel to
kernel-smp-2.4.21-15.EL, installed GFS-6.0.0-1.2 and GFS-modules-smp-6.0.0-1.2
from rpms, perl-net-telnet, i run ntpd.
Following the documentation the modules are loaded, wrote pool-configfiles for
one filesystem and the pool for the cca. pool_tool -c enables the pools as it
should.

According to the docs i should now be able to pool_assemble the pools on all
three nodes, however: that works for me only on the first node where
pool_tool -c was run.
On the other 2 nodes i just get pool_assemble zu succeed after pool_tool -c'ing
the two pool-configfiles there.
After creating the clusterconfig ( {cluster,fence,nodes}.ccs ) i can write that
config to the cca-pool with 'ccs_tool create'. That config from the cca-pool
can now only be extracted/read again properly from the node that wrote it.
I can follow the instructions and finally create/mount the gfs on each node,
but its just local and not spread on all three nodes.

What am i doing wrong?

thanks, and a happy new year, Christian.