I sometimes end up in a situation where I may lose the kernel that contains the KubeCluster object used to create the Dask Cluster.
Is there a way to delete an existing KubeCluster from another kernel?
Alternately, if not in the API, what commands can delete it from inside the Kubernetes cluster itself?
I tried deleting all pods and the service, but think I may be missing something, because when I try to spawn a fresh DaskCluster, I get errors on init
Hi @tszumowski
I have been using KubeCluster recently, not from the a (jupyter) kernel though.
From my experience, I was able to delete pods manually in case I quit the script that spawned the cluster. From what I understand - each initiation of a client + KubeCluster is standalone - so it seems weird you have an issue - maybe it has to do with not having enough resources in the cluster?
As for deleting from the API - did you try to connect the re-connect the client and then client.cluster.close()
it? I have no idea if it will work but worth a try
1 Like