Open
Description
Problem
It seems like PrometheusHandler
is consuming CPU even when there is no activity. More discussion in discourse thread.
We are setting up a PeriodicCallback for this handler which is eventually consuming that CPU. The not so clear part is that this handler is not registered with any endpoint for Prometheus to scrape the data. So, maybe this is vestigial piece of code?
Proposed Solution
If this PrometheusHandler
is not critical, maybe we can set default config of enable_prometheus_metrics
to False
.
If PrometheusHandler
is indeed unusable in the current state, we need to fix it for Prometheus to be able to scrap the metric data.