As with many things, I think the devil in that quote is in the details.
Failover domains are not required for the cluster to run *at all*, but I
think they may be required for the cluster to behave as you'd like.
For example, if I wanted to have multiple front-end nodes serving a GFS
filesystem, with a load balancer out in front of the cluster to redirect
requests, I wouldn't need a failover domain. Every node would be mounting
the GFS filesystem, so there wouldn't be a service I'd need to fail over to
anywhere. However, if I wanted to run a web server on one node and have it
migrate to another node if the first one failed, then I would need a
failover domain.
Regards,
James Chamberlain
On Thu, 29 Nov 2007, Scott Becker wrote:
jr wrote:
from my understanding a failover domain is required whenever you want
other nodes to take over a service. the subset is if you make it
restricted, isn't it?
regards,
johanne
From:
https://www.redhat.com/docs/manuals/enterprise/RHEL-5-manual/en-US/RHEL510/Cluster_Administration/s1-config-failover-domain-conga-CA.html
"Failover domains are /not/ required for operation."
scottb
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster
