I'd need to force moving a Pod from one Openshift Node to another, in order to do some performance tests. From the documentation it seems that setting the nodeSelector in the Deployment config is the way to go, but it doesn't work, according to my tests. Here is what I have tried as a test:
Create nginx Pod
oc new-app -f https://raw.githubusercontent.com/sclorg/nginx-ex/master/openshift/templates/nginx.jsonThe Pod is running on "ip-10-0-121-229.us-east-2.compute.internal" Node. Now I patch the node selector, setting a target Node:
oc patch dc nginx-example -p '{"spec":{"template":{"spec":{"nodeSelector":{"kubernetes.io/hostname": "ip-10-0-169-74.us-east-2.compute.internal"}}}}}'However the Pod is still running on the same Node. Even by killing the Pod it re-attaches there. Any suggestion? Thanjs
You can use nodeName field in your pod spec to schedule the pod yourself, rather than scheduler do it.
oc explain pod.spec.nodeName
FIELD: nodeName <string>
DESCRIPTION:
NodeName is a request to schedule this pod onto a specific node. If it is
non-empty, the scheduler simply schedules this pod onto that node, assuming
that it fits resource requirements.you can patch it in similar way. Don't forget to delete nodeSelctor fields
oc patch dc nginx-example -p '{"spec":{"template":{"spec":{"nodeName": "ip-10-0-169-74.us-east-2.compute.internal"}}}}'
you can try this -
kubectl get pod -o widethis will give you the VM on which your pod is running
then execute
kubectl cordon {name_of_that_node_in_which_POD_is_running}then delete the pods, those you want to get aligned to other node
then run
kubectl uncordon {the_node_that_was_cordoned_above}