Den 2014-03-10 15:51, john.r.rothlisberger at accenture.com skrev:
This has been bugging me for years. Let's assume you ack an alert for the service "W3SVC" for 4 hours for maintenance, which you have stopped on purpose for whatever reason. Let's assume you also monitor services for SQL, Java, etc. If one of those also fails during the 4 hours of ack time - you won't get an alert. There are 3 tests that can have multiple components which I would like to know if new alerts arrive while a test has been ack'd: disk, procs, & svcs.
As I see it, the core of the problem with this is that Xymon currently bundles tests by the method in which they are tested, not by the "thing" that they test.
So you have a "procs" status containing the status of multiple processes. But these processes may not have anything to do with each other, so handling alerts and acks based on this combination-status causes problems.
Same issue with e.g. "disk" or "http" status.
As the old-timers here know, the reason for this is historical. That's a bad excuse, though, and it is something that needs changing. It is "in the pipeline"; since we have all of the rules for checking e.g. processes defined in analysis.cfg, we can also tell Xymon to put the analysis result in a different column than the "procs" column.
I just need to write the code to do that :-(
Regards, Henrik