Runtime instrumentation

[morigs]

Is your instrumentation request related to a problem? Please describe

It's very useful to have metrics about platform runtime - heap status, GC metrics, event loop blocks etc...
OpenTelemetry Go provides runtime instrumentation for this.

For example, if application code behaves strangely, these metrics can understand if this because of GC lock or because of event loop block.

This instrumentation can move OTel forward to full-featured APM solution.

Is it applicable for Node or Browser or both?

Mostly Node

Do you expect this instrumentation to be commonly used?

I believe, it's very useful. prom-client includes these metrics and they're commonly used

[tomerfriedman]

Going to start working on this (first contribution). Any objections?

I already implemented something like that on my own. And was thinking to put it under "packages" here. (same as host-metrics)

[vmarchaud]

Any objections?

None, I will assign the issue to you !

And was thinking to put it under "packages" here. (same as host-metrics)

Yeah i think its the most logical place

[dyladan]

Maybe this should go in the core repo? Similar to http and grpc instrumentation I think it is quite important and it is likely to get more maintainer attention there

[dyladan]

Host metrics too maybe

[daniel-white]

would love to see the GC metrics, something along the lines of https://github.com/SimenB/node-prometheus-gc-stats

[LiranBri]

+1
seeing event-loop delays is super required to be able to monitor the health of a nodejs service

[pikalovArtemN]

Add some prom-client metrics
#2136

[david-luna]

Hi 👋 !

thanks to @maryliag we have some metrics in semantic conventions :)
open-telemetry/semantic-conventions#991

I was about to start adding these into runtime metrics instrumentation but I found this issue and PR. I wonder if it makes sense to have a dedicated PR for the already added metrics (could be a partial of the ongoing #2136) or it's preferable to update & continue #2136 although it adds some metrics not defined yet in semconv

[pikalovArtemN]

Hi 👋 !

thanks to @maryliag we have some metrics in semantic conventions :) open-telemetry/semantic-conventions#991

I was about to start adding these into runtime metrics instrumentation but I found this issue and PR. I wonder if it makes sense to have a dedicated PR for the already added metrics (could be a partial of the ongoing #2136) or it's preferable to update & continue #2136 although it adds some metrics not defined yet in semconv

I fix up all today with convention

[daniel-white]

Shouldn’t this now go in the primary repo since it’s based on the conventions?