[ClusterLabs] Running Pacemaker/Corosync in HA pair on stateless machines, why not?

[Jeff Johnson]

Greetings,

I've configured HA clusters for some time and I find myself wanting to 
do this on stateless machines (PXE booted RAMdisk OS image). I hear that 
you cannot run Pacemaker/Corosync on stateless pairs but I haven't heard 
exactly why.

It seems to me that if I have an OS image (PXE/RAMdisk) that contains 
all the proper packages, a shared corosync key and config, and a fully 
configured cib file in the proper place I should be able to boot two 
independent machines off of that image with the configured IPs and 
hostnames and get a quorum just like a stateful pair booting up.

If one stateless node gets fenced, the survivor has the current CIB and 
resources, continues to function. If the fenced node reboots it has the 
original cib and when it rejoins the cluster the surviving node updates 
the cib on the newly joined node. If both get fenced or otherwise 
powered down, they both reboot with the original cib that defines 
resources and locations.

What am I missing? This is for a simple two node HA cluster.

Thanks...

-- 
------------------------------
Jeff Johnson
Co-Founder
Aeon Computing

jeff.johnson at aeoncomputing.com
www.aeoncomputing.com
t: 858-412-3810 x1001   f: 858-412-3845
m: 619-204-9061

4170 Morena Boulevard, Suite C - San Diego, CA 92117

High-Performance Computing / Lustre Filesystems / Scale-out Storage