[Linux-cluster] FS mount point must really be unique ?

[Lon Hohberger]

On Tue, 2007-11-27 at 06:40 -0800, Roger Peña wrote:
> Hi
> 
> I am getting a resource confict, because identical
> mount point definitions in Global FS resources, I read
> the FAQ:
> http://sources.redhat.com/cluster/faq.html#rgm_wontstart
> and it clearly say that:
> "The mount point is also the same, and that must be
> unique."
> 
> ok, my question is:
> is there any way I can avoid that check ?.

Depends - what does your cluster.conf look like?

> My services are email services, and my common FS
> resource is the email queue but I can imagine that two
> differents web services with the same /var/www mount
> point with different content are perfectly valid
> cases.

If you're using GFS, you can reference the same file system from
different services.

If you're using ext3, you can disable the check by
changing /usr/share/cluster/fs.sh.  Change the 'unique' attr of the
'mountpoint' parameter to 0 on all nodes.  Delete the backup file when
you're done (or at least chmod -x...).

After that, increment the version # in cluster.conf and run 'ccs_tool
update /etc/cluster/cluster.conf'.  This will cause rgmanager to reread
the configuration and all resource agents.

With unique="0", attribute collisions are ignored for that parameter.

You can check using 'rg_test test /etc/cluster/cluster.conf'

-- Lon