[Linux-cluster] GFS deployment questions

Alan A alan.zg at gmail.com
Mon Sep 15 19:20:58 UTC 2008

I am trying to setup cluster test environment mainly for the GFS. The
application that will be supported likes to take control of CPU thread and
is not clusterable in terms of process sharing, however it does talk to
SAN where GFS comes in. This is not something we did before so reading
manuals and this email list is our primary source of information. I have few
simple/ or not so simple questions, and would be grateful for any pointers
in the right direction.

Going through documentation, it appears that after fencing is setup, one
would need to allocate resources to support cluster. Is this only for High
Availability Cluster or any cluster, meaning - if I am only to test GFS do I
need to allocate resources in order to launch GFS.

I know that GFS service must be launched on a cluster. Once fencing is
configured, what are the steps required to deploy GFS, and any potential
'bumps' to watch out for?

Some questions regarding fencing:
We are using HP DL385s as the test model. After researching possible fencing
devices we decided to try HP ILO and SCSI fencing before testing HBA switch
fencing. For HP ILO fencing - each of the nodes has HP ILO name, but I am
not sure what goes into hostname - is it the name of the node, or the IP
address of HP ILO?
There isn't a lot of information about SCSI fencing, I would appreciate any
pointers for/against SCSI fencing.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20080915/8f638b52/attachment.htm>

More information about the Linux-cluster mailing list