On 7/24/2017 10:19 AM, John Tullis wrote:
I'm getting an extremely strange behavior with alerts.
Several recipients are complaining that they are getting alerts for servers that they don't manage and aren't assigned. There are about 200 lines in my alerts.cfg file.
I've verified that the alert rules are accurate and running the test alert script (./bin/xymoncmd xymond_alert --test sever.name cpu) shows the correct, assigned user and the info page for the servers show the right user's email.
Anyone know what I'm missing? The same problems happened with my old Hobbit server as well so it could be a config file issue but I've looked thoroughly and haven't been able to track it down yet.
Hi,
I'd start by adding the --cfid option to xymond_alert in tasks.cfg. This will cause the alert rule in question to be printed along with each alert that goes out, which usually helps narrow down what's going on. If you have a alerts.cfg file that spans multiple actual files (via include statements), you can run xymoncmd xymond_alert --dump-config to see what config lines line up with what rules. (Keep in mind that config line #'s will change once you do start making changes to things.)
If that doesn't help, or just points you to catch-all rules, you can run xymoncmd xymond_alert --test <host> <test> [other options] to trace down exactly what evaluations are occurring, although with a large alert config file you'll get a correspondingly large amount of output.
You'll also want to make sure you're running 4.3.25 or higher, as that fixed a few bugs involving host records in xymond_alert.
HTH, -jc