USE CASE:

Foreman telemetry stack tracks the performance of a system based on its current runtime. If there were a puppet metric, it would be dramatically simpler to correlate performance changes with system configuration changes.

Technical requirements:

Can a PCP plugin be written for puppet client that tracks the simple results of puppet runs on the system:

- Applied
- Restarted
- Failed
- Restart Failures
- Skipped
- Pending

I'd expect these metrics to only tick up at the conclusion of the puppet run (and mostly just pull them in from the post run status report) and to be zero for all other times.

ANALYSIS:

We can add a new metric that will track number of resources with label of "state". Each report will increase those counters, so monitoring will be able to show rate of resources applied/skipped etc.