Wonder why not use JMX collect for JVM metrics.

[Roiocam]

Motivation

I am currently changing my application metrics from JMX to OTEL. Before I used Prometheus-JMX-exporter, they supported JVM metrics and Grafana Dashboard.

I don't see anything like this on OTEL-JAVA, but using the same way to collect JMX like Prometheus-JMX-exporter also works on OTEL.

I could use Meter.batchCallback to start JMX-Scraper work, on that small demo, everything is working fine.

JMX has a ton of metrics for JVM, I believe it could help OTEL JVM metrics complete.

[Roiocam]

Sorry for that, I notice there is already the same thing on here: https://github.com/open-telemetry/opentelemetry-java-instrumentation/tree/main/instrumentation/runtime-metrics/library/src/main/java/io/opentelemetry/instrumentation/runtimemetrics

[mateuszrzeszutek]

Hey @Roiocam ,

As you've already noticed, we are in fact using JMX to get the JVM metrics: https://github.com/open-telemetry/opentelemetry-java-instrumentation/tree/main/instrumentation/runtime-metrics/library/src/main/java/io/opentelemetry/instrumentation/runtimemetrics
We're slowly adding new JVM metrics, but since they need to be specced it unfortunately takes time.

If you want to listen for custom JMX metrics (as defined in a config file, for example), we have an open issue for this: #6131

[Roiocam]

Thanks for your time.

I noticed this is not completed runtime metrics, something is missing like memory metrics(not memory pool), os physical memory, os available processors.

does anyone work on this? tell me if not and then I could support this.

[trask]

hi @Roiocam, we would ideally like to get new metrics added to the specification first, check out https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/metrics/semantic_conventions/runtime-environment-metrics.md#jvm-metrics

[Roiocam]

Thanks for your advice, I will enhance the specification first by opening an issue.