Trying to move workloads from on-demand instances to spot instances on a test cluster with guestbook installed with 20 frontend replicas.
The rescheduler fails to move pods with the following error message:
k8s-spot-rescheduler-6b85d59bf5-gxlm7 k8s-spot-rescheduler E1004 11:37:47.920923 1 rescheduler.go:256] Failed to get pods for consideration: staging/frontend-69859f6796-jx7zs is not replicated
k8s-spot-rescheduler-6b85d59bf5-gxlm7 k8s-spot-rescheduler E1004 11:37:47.920929 1 rescheduler.go:256] Failed to get pods for consideration: staging/frontend-69859f6796-ppz8p is not replicated
k8s-spot-rescheduler-6b85d59bf5-gxlm7 k8s-spot-rescheduler E1004 11:37:47.920935 1 rescheduler.go:256] Failed to get pods for consideration: staging/frontend-69859f6796-fpzp6 is not replicated
adding the flag --delete-non-replicated-pods=true solves this but that is not the right way.
Any suggestions on why this happens?
Also is there any plans to support eviction of pods that contain a specific label? This would help to do a lot of controlled eviction