Manage beyla metrics retention period while using prometheus http endpoint.

[siba-prosimo]

trafficstars

Hi Team, We are using promtheus http endpoint to read and ship beyla metrics to our centralised tsdb, so far we have just enabled instrumentation on a single app and the memory usage for beyla demonset is around 500mb but it would mostly shoot up once we enable it for multiple apps. Considering our promethues scrap frequency of 30 seconds, it would be great if we have a way to manage metrics retention period at the endpoint.

[grcevski]

Hi there @siba-prosimo, what the programming language of the applications you are monitoring? We are working on reducing the memory consumption and we merged a PR in main today related to this. We are still investigating more here.

[siba-prosimo]

@grcevski We have currently instrumented an nginx based app, however most of our apps are using golang.

[marctc]

I believe that thanks to metrics expirer + memory optimizations this shouldn't be a problem anymore. Feel free to reopen if problem still persist.