Hey all,
Reading from https://twitter.github.io/finagle/guide/Metrics.html, is jvm:mem:current:used metric (prometheus) the definitive metric to monitor for linkerd's memory usage? Or, are there other metrics which collectively can tell us how much memory linkerd is using?
Thanks for the suggestions.
Best Wishes,
Amit.
Hi @amitsaha. jvm:mem:current:used is a valid metric to use, also recommend looking at jvm:heap:used.
We have an example linkerd health dashboard that provides a good view into linkerd’s memory usage, it’s part of linkerd-viz, have a look at the specific dashboard at: https://github.com/linkerd/linkerd-viz/blob/master/linkerd-health-dashboard.json
Thanks for the link to the dashboard. Deconstructing from the JSON, I think (jvm:heap:used /jvm:heap:max )*100 is a fair metric to monitor as the percentage of memory used by linkerd?
I am a bit confused about the jvm:current:used metric. Is this in addition to the heap memory used? Thanks in advance.
Correct, (jvm:heap:used /jvm:heap:max )*100 is a good measure of percentage memory used.
jvm:current:used = jvm/heap/used + jvm/nonheap/used
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.