[Linux-cluster] Configuration of a 2 node HA cluster with gfs

[birger]

I think I sent this with a wrong sender address, as I didn't see it 
apear on the list. I have solved the problem with starting ccsd in my 
original message (if it ever went out)

I have just installed the cluster package, and I am now looking for some
help on how to use it :-)

I have a lot of experience with Veritas FirstWatch and some with
SunCluster, so I am not new to HA services. Now I have this server that
I have to get up and running as quickly as possible... And I find little
documentation about how to get this software up and running from a fresh
install.

I have one old file server with external scsi disks, and one new server
with a SCSI-attached Nexsan ATAboy RAID array.

I want to set up the new file server as half of a 2-node cluster and get
it into production. Then move over data (and disks) from the old server
until I can reinstall that one as the second cluster node.

I first thought along the lines I am used to from Solaris and
clustering. I wanted to set up 2 services: 1 NFS service that would take
the disk with it when it moved, and 1 samba service that NFS-mounted the
disk from the nfs service.

After looking at the redhat stuff I am thinking:
- Mount the disks permanently on both nodes using gfs (less chance of
nuking the file systems because of a split-brain)
- Perhaps also run NFS services permanently on both nodes, failing over
only the IP address of the official NFS service. Should make failover
even faster, but are there pitfalls to running multiple NFS servers off
the same gfs file system? In addition to failing over the IP address, I
would have to look into how to take along NFS file locks when doing a
takeover.
- samba running as a service that fails over if a node goes down.

Can anyone 'talk me through' the steps needed to get this up and running?

First attempts at starting ccsd failed with
Failed to connect to cluster manager.
Hint: Magma plugins are not in the right spot.

I fixed this by cd'ing down into magma in the cluster directory i 
fetched with cvs and doing
make clean
make
make install

When I did a make from top-level of the cvs sources magma got built with 
plugin dir pointing into the source directory. Just recompiling (without 
rerunning configure) fixed it. Something to look into for the maintainers?

I now have my first gfs file system, but I get a permission denied when 
trying to mount it. How should I diagnose this?

-- 
birger