[ClusterLabs] FEEDBACK WANTED: possible deprecation of nagios-class resources

[Ken Gaillot]

Hi all,

I am considering deprecating Pacemaker's support for nagios-class
resources.

This has nothing to do with nagios monitoring of a Pacemaker cluster,
which would be unaffected. This is about Pacemaker's ability to use
nagios plugin scripts as a type of resource.

Nagios-class resources act as a sort of proxy monitor for another
resource. You configure the main resource (typically a VM or container)
as normal, then configure a nagios:<PLUGIN> resource to monitor it. If
the nagios monitor fails, the main resource is restarted.

Most of that use case is now better covered by Pacemaker Remote and
bundles. The only advantage of nagios-class resources these days is if
you have a VM or container image that you can't modify (to add
Pacemaker Remote). If we deprecate nagios-class resources, the
preferred alternative would be to write a custom OCF agent with the
monitoring you want (which can even call a nagios plugin).

Does anyone here use nagios-class resources? If it's actively being
used, I'm willing to keep it around. But if there's no demand, we'd
rather not have to maintain that (poorly tested) code forever.
-- 
Ken Gaillot <kgaillot at redhat.com>