We're in the process of setting up OpenNMS on most of the linux machines in our infrastructure and just started to implement the various metrics that come out of Apache when ExtendedStatus is set to "on". We starting pointing OpenNMS at one of our cloud instances which - other than having apache and getting (literally) no requests ATM - is doing nothing; this won't be the case shortly but we wanted to get everything setup beforehand.
I'll spare everyone the details but I will say that we were receiving data just fine and then all of the sudden it stopped and nothing was coming in. After some digging it turns out that OpenNMS was throwing an unhandled exception b/c it was expecting a regex that was expecting CPULoad was not getting it when it tried to poll http://127.0.0.1/server-status?auto.
We stirred things up a bit on the machine to get the (apache) cpu load going and it is now showing up. IMHO, this seems a little counter-intuitive. If I am polling something and it doesn't report it, to me that means something is broken. For example, I would be just as shocked to run a "top" on this machine - when it is not doing anything - and have it quit out of the command before showing me anything.
Is this the expected behavior for Extendedstatus?