Add cgroups CPU quota and throttling metrics - #1039
Conversation
Add metrics related to CPU quotas and CPU throttling (Linux CFS bandwidth control), as well as the total CPU usage from Linux cgroups CPU accounting. Those metrics can be useful in multi-tenant cloud environments, in particular on Elastic Cloud nodes that use CPU boosting (vCPU credits). Signed-off-by: Michal Kapalka <mkapalka@adobe.com>
ad384a0 to
56195f7
Compare
|
@SuperQ @sysadmind it would be great if you could have a look at this PR and tell me if there's anything missing that I should add. Thanks in advance! |
|
Update: we have been using this branch successfully in production for quite some time now and those new metrics are very helpful, maybe it's worth merging this PR to make it easier for others to benefit from this as well? @SuperQ @sysadmind |
sysadmind
left a comment
sysadmind
left a comment
There was a problem hiding this comment.
I'm not sure why this PR hasn't triggered the CI process. You may need to rebase, or when you update this PR that might trigger the CI to run.
| }, | ||
| Labels: defaultNodeLabelValues, | ||
| }, | ||
| { |
There was a problem hiding this comment.
I think these metrics should be converted to seconds. It's typical practice for prometheus metrics to always be in base units - https://prometheus.io/docs/practices/naming/#metric-names
2 participants
Add metrics related to CPU quotas and CPU throttling (Linux CFS bandwidth control), as well as the total CPU usage from Linux cgroups CPU accounting. Those metrics can be useful in multi-tenant cloud environments, in particular on Elastic Cloud nodes that use CPU boosting (vCPU credits).