Marcin Kulik
Marcin Kulik
I can observe similar behavior on asciinema.org: 
Thanks. Yeah, the problem is definitely there. I'm trying to pinpoint it now.
I think it's the built-in prometheus endpoint (`http://IP:9568/metrics`), which when not queried, accumulates aggregated data in ETS table.
That's likely it: https://github.com/beam-telemetry/telemetry_metrics_prometheus_core/issues/52
I nailed it. This is mem graph from asciinema.org, with the prometheus endpoint disabled. Note the last quarter - flat as a table :) 
I'll release a new version with the fix soon.
The built-in prometheus endpoint provided the following stats: https://github.com/asciinema/asciinema-server/blob/b3a852d769da24b1360562b443e30ca41de23351/lib/asciinema/telemetry.ex#L48-L76 - so, BEAM VM mem/cpu stats, HTTP request stats, database query stats and background job stats. It didn't have application-level stats...
> Yeah it was weird for me when I saw it in the logs (admin panel listening on 4002) while docs said there's no admin panel. Both the dashboard and...
@jiriks74 this one runs on its own port, 9568.
@jiriks74 the new release is out (docker image), I'm just wrapping up the release notes.