Usage of jvm_memory_pool_allocated_bytes_total mertics in Grafana

97 views

Skip to first unread message

Denis Voloshin

unread,

Apr 18, 2020, 7:45:17 AM4/18/20

to Prometheus Users

Hello.

I need some help in getting how jvm_memory_pool_allocated_bytes_total should be used.

I have EKS cluster which runs a few based JVM deployments.
Per my understanding jvm_memory_pool_allocated_bytes_total metric provides the current total allocated mem usage by pool type.
To get the total JVM mem footprint I need to sum all of them.
Now to get it in Grafana

My Grafana query looks like

sum(jvm_memory_pool_allocated_bytes_total{instance=~"$instance", service=~"$service", pool=~".*"}) by (instance)

The result I got doesn't make sense.
What I'm getting is

allocated 172.31.66.235:8084 5.071 TiB

Notice my JVM is started within docker using the following JVM args

ENTRYPOINT ["java", "-XX:+UnlockExperimentalVMOptions", " XX:+UseCGroupMemoryLimitForHeap", "-jar", "my-jar-1.0.jar"]