https://rancher.com/ logo
#rke2
Title
# rke2
b

billowy-midnight-87589

03/01/2023, 9:21 AM
Hello, I created RKE2 cluster
v1.24.8+rke2r1
3x VM’s for master nodes [control plain] and 2x physical server as worker nodes. It works fine and run workload well. Most of workload if ArgoCD workflows it runs do some work and finish pods. But I have issue that those workloads not balanced between worker nodes (physical server) they runs mostly on one worker. How I can balance workload between worker nodes? Average workload run time is about 2min.
c

creamy-pencil-82913

03/01/2023, 9:30 AM
Kubernetes makes no attempt to balance node load. If a node has sufficient resources to host a pod, it will be scheduled there. That is just how the scheduler works. You should read the docs at https://kubernetes.io/docs/concepts/scheduling-eviction/topology-spread-constraints/ for possible ways to get closer to what you want.
b

billowy-midnight-87589

03/01/2023, 9:54 AM
yes that I saw documentation
most of this is configuration on pod’s manifest, but there is maybe some configuration in RKE2 server side?
c

creamy-pencil-82913

03/01/2023, 4:12 PM
No. If you want the scheduler to spread your pods out, you need to follow those docs and tell it how you would like them spread.
3 Views