We are building a ONE env.
Something that isnt really clear to me is when doing a HA setup with the RHEL clustering tools as described in the manual whether one should sync/cluster the FS underlying the cluster nodes (the entire disks or the /var/one dir)
Because what makes me wonder is that if a configuration change is done on one of the nodes this would not apply for both the machines which can cause problems, so to me it seems that either this setup would require ‘double’ administration or the files will have to be synced in one way or another.
Yes you need to keep all the master nodes in sync in terms of configuration files. Those are pretty static, though. The actual layout of the system is on the DB, i.e. number of hosts, images etc,
Is this commonly done manual or over a shared FS(GFS/cLVM etc)?
What I am trying to prevent is more software, more points of failure and more things to maintain, therefor i’d rather not introduce more stuff basically.
What we do is configure one node and verify the installation, then we clone
the configuration. When the configuration is changed we trigger the
synchronization manually.