excessive legends for metric cpu

[chenatu]

excessive legends for metric cpu

[anryko]

That's because you have 8 cores on your machine and by default collectd is pushing a metric for each individual core.
If you have collectd version >= 5.5, you should use this configuration: https://github.com/anryko/grafana-influx-dashboard/blob/master/examples/collectd.conf.d/cpu.conf
Otherwise, for older collectd version I would suggest: https://github.com/anryko/cpu-collectd-plugin

[anryko]

_.uniq behavior changed in new version of lodash, which introduced a bug after migration to Grafana4.
Please check out the last commit and close the issue if problem is solved.

[anryko]

If you want to graph default unaggregated collectd cpu metrics per core, you should apply derivative(sum) function to each cpu graph in getdash.conf.js:

  plugins.cpu.cpu = {
    'graph': {
      'system': {
        'color': '#EAB839',
        'apply': 'derivative(sum)',
        'alias': '@description'
      },
...

It will work OK, just a bit slower, because for your 8 core setup you will work with 8*8 metrics instead of just 8 if those are aggregated beforehand.

[chenatu]

Thank you, it solves my problem