Closed zhuoyuan-liu closed 1 month ago
Thank you for the PR! I will add some comments in a code review 🙂
I have updated the flag. Should I start to clean up the old metrics components for osctrl-tls?
I have updated the flag. Should I start to clean up the old metrics components for osctrl-tls?
You mean the emitted metrics? Instead of removing them, it would be better if we can find a way for those to be utilized by prometheus too.
@javuto I have added the flag, could you please take another look?
@javuto I Just fixed the conflicts.
This PR added Prometheus metrics support for osctrl-tls. Issue: https://github.com/jmpsec/osctrl/issues/504
Currently, it has all default metrics go metrics such as memory info and gc duration etc. Also, we added one custom histogram metric to measure the total request/latency to each osquery endpoint.
The metric server would run as a separate service in a different port. We can query e.g. P90 latency for each endpoint with different environments with the histogram metric.
Also the request rate for different path and status code.