Extended metrics
We want metrics for:
-
histogram call times for each endpoint -
python interpreter data (gc runs / memory ussage) (pid label) -
ratelimit events
Special care musst be take to merge those over uwsgi processes / threads. We could look how prometheus_flask_exporter
does that, or create our own solution via shared memory.
This will probably result in a mandatory dependency on prometheus-client
, that's why it's targeting 3.0.0.