On Tue, 2011-12-20 at 15:39 -0600, Al Amyot wrote:
If I understand alert definitions correctly, they can only test availability, traits and metrics within one resource. This means that I cannot test for server1=DOWN AND server2=DOWN as a simple example. Is this correct?
Yes, that is correct.
If so, has any consideration been given to making the alerting system more ‘global’ in scope?
Yes, see https://bugzilla.redhat.com/show_bug.cgi?id=725908
This actually just came up again a few days ago. It is something people have asked for - it will need a large change to the way the alert subsystem works so it isn't trivial to implement. Again, nothing currently in the works. If anyone wants to contribute and come up with a design/implementation that provides this feature, they are welcome to join in the fun :)