-
Story
-
Resolution: Done
-
Undefined
-
None
-
1
-
False
-
None
-
False
-
-
-
RHOBS Sprint 20, MCO Sprint 20, MCO Sprint 21, MCO Sprint 22
-
No
In several of the ACM dashboard widgets, the following metric is used:
avg(rate(node_cpu_seconds_total{cluster=~"(...)",mode="idle"}[$__rate_interval]))
this has a cardinality of CPUs*cluster and can therefore be a lot of data to fetch when having a large amount of managed clusters, causing poor dashboard performance.
Make sure a recording rule is available in order to improve dashboard loading times.
Acceptance criteria:
- Recording for CPU usage is available for all managed clusters