How to graph CPU IO Wait times?

Feature Requests
1

None of my Linux servers are showing IO Wait in the detailed CPU usage graphs. They’re only showing Nice, Idle, System, and User. But, doing searches through the ucd-mib.php and related files, there’s lots of mentions of iowait. It just doesn’t appear in the graphs.

What am I missing? How do I get the iowait added to the detailed CPU graphs? It’s polled via SNMP, just not graphed anywhere. If I click the “view rrd command” link on the graph page, neither iowait nor wait appear anywhere in the command.

There’s one specific server that I would really like to get this enabled on, if there’s something I have to do manually. Otherwise, if there’s a way to enable it for all systems, that would be great.

2 Likes
2

I second that request, this is really a big thing missing in LibreNMS, imoho. There’s neither a facility to alert for high CPU wait, or high disk IO.

3

Just found this, maybe it could work, coupled with mk_agent? Not sure yet.

https://exchange.nagios.org/directory/Plugins/Operating-Systems/Linux/check_cpu-2Epy--2D-usage-and-IO-Wait/details

4

The code to collect it is 9 years old which dates back to the Observium days. So I guess it was never done then, and nobody since has done it either.

The fix is to add the graphs to includes/html/graphs/device/ucd_cpu.inc.php I guess