SOLVED: Can someone spot the obvious error (because I c.....

[adamnealis]

To start to get the hang of cacti I've written a short perl script to simulate classic top(1) style CPU usage. And I want to graph the results.

I've got a slight problem. The fourth value "idle" is ignored by the poller.php job, as can be seem by the -t argument to rrdtool update below.

I'm convinced I've missed out a simple step, but I just cannot find out what!

If someone could point out my obvious error I'd be very grateful!

Thanks,
Adam.

Here's example perl script output:

sys:4.13163135099275 user:34.27011344309 wait:38.6717984087562 idle:22.9264567971611

Here's the corresponding line from cacti.log:

01/22/2006 03:50:01 PM - POLLER: Poller[0] CACTI2RRD: /usr/local/bin/rrdtool update /usr/local/share/cacti/rra/sys_22.rrd --template sys:user:wait 1137945000:60.310830602339:36.6894968275809:1.16954997910504

Here are the respective screens showing how this DS is configured and the perl script.

[adamnealis]

Don't know exactly what was wrong, but I first reordered the output fields in my script, and still the "idle" DS was ignored.

So I removed "idle" from the data input method.

This broke the graph management bit.

When I re-created the "idle" data source I noticed that cacti began to correctly collect user, sys, idle and wait.

[gandalf]

Please pay attention to print the output line as a single statement (no iterative prints) as the cacti poller will not wait if there might be another print if he's already got a first one ...
Reinhard