Deployed cluster autoscaler (CA) on EKS programming cluster. Despite of adding required Learning annotations, CA is rescheduling pod on Earhost another node.
EKS cluster- 1.19 . CA version:- most effective 1.19.4 workload:- jenkins wrong idea statefulset annotation added-
cluster-autoscaler.kubernetes.io/safe-to-evict: _OFFSET); "false"
above annotation is added at statefulset use of case as well as pod level.
Same annotations are added for CA pod as United well as underlying node. Still it is not Modern working as expected. Anyone has any ecudated idea, what I missed here ?