* Frequency usage percentage stats are by kHz instead of Hz.
Correct the labels. (Linux natively uses kHz for CPU frequency stats,
but collectd scales the current frequency stats item to Hz.)
* Show frequency usage percentage graph before the transition counts,
as it is more descriptive.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
* Use the new data series sorting and coloring options to
display the frequencies in order and with matching coloring in
different cores.
* Fix the y-axis text and legend in the frequency usage graph
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Collectd 5.9.0 changed the data structure of the cpufreq plugin:
CPU cores are now handled as separate plugin instances.
There are also new data items per core:
* time spent at each frequency
* amount of frequency transitions
Enable these new data items, but initially hide them behind
a new config option "ExtraItems" (default: disabled), as
the amount of graphs in multi-core systems could be rather large.
Note that the frequencies are not (yet) sorted, so the
information value of the time-spent graph is semi-random.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
The original data model definition assumed data from a quad-core CPU,
which caused errors with single- and dual-core processors.
Adjust the data model to work with also them.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Add support for 'cpufreq' plugin in collectd, which monitors
the CPU cores' frequencies. Some platforms enable dynamic
frequency scaling according to CPU load.
'cpufreq' plugin can currently be built for x86 and mvebu.
(and it should likely be enabled also for ipq806x)
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
