Enable collectd/cpu metrics as default? about signalfx-agent HOT 3 CLOSED

Yes, that is essentially a bug given our criteria for which metrics get turned on by default or not. Any metric used in built in content is supposed to be on by default.

We'll switch it so it gets categorized as a "host" metric so it won't count against your custom metric count, and also update the agent to make it default out of the box. In the meantime, you can enable it like this:

- type: collectd/cpu
  extraMetrics:
    - cpu.idle

from signalfx-agent.

Thanks! Already have.

Any chance of having the rest of the cpu metrics enabled by default? I don't believe it's currently possible to tell if the host is spending all it's time in iowait, or in the hypervisor, or anywhere else. CPU utilization alone can be misleading. Those metrics can be immensely useful and it seems odd that they count against the custom metric count in general. In general, other products we've used collect them by default. Apologies if that's a bigger discussion.

from signalfx-agent.

Yeah when we switched to the host-based pricing model the determining factor for what was host vs custom was whether it is in the built-in content that we had, and in all of that only cpu.utilization is used (and cpu.idle which should now be default as of 4.12.0, and should also be categorized as host). I would recommend bringing it up with your account exec.

from signalfx-agent.