--
You received this message because you are subscribed to the Google Groups "Prometheus Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-developers+unsub...@googlegroups.com.
To post to this group, send email to prometheus-developers@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-developers/e36ab2d5-c41b-4fdb-b772-14ac9ad54e58%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
They come from https://en.wikipedia.org/wiki/Load_(computing). They're not very useful, as it's only an indicator of the run queue. But we include it for completeness, and it can sometimes be used as an indicator of problems. For example, many people traditionally use it to indicate IO problems, but node_cpu{mode="iowait"} is much more useful for this.
On Sun, Dec 10, 2017 at 7:24 AM, tilak kumar <tilakk...@gmail.com> wrote:
Hi,
What does metrics node_load1, node_load5 and node_load15 mean and how is it different from node_cpu ? Could not find documentation on these.
Thanks
Tilak
--
You received this message because you are subscribed to the Google Groups "Prometheus Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-developers+unsub...@googlegroups.com.
To post to this group, send email to prometheus...@googlegroups.com.