Zhankun Tang created SUBMARINE-644:
--------------------------------------
Summary: Helm charts deletion won't delete CRD and may have CRD
object leaked
Key: SUBMARINE-644
URL: https://issues.apache.org/jira/browse/SUBMARINE-644
Project: Apache Submarine
Issue Type: Improvement
Components: Backend Server
Reporter: Zhankun Tang
There're a bunch of issues in helm charts found:
1. If the user has CRD "tfjobs.kubeflow.org",
"podgroups.scheduling.incubator.k8s.io", "pytorchjobs.kubeflow.org" already
exists in K8s cluster. It will report below errors:
{code:java}
$ helm install submarine ./helm-charts/submarine
Error: rendered manifests contain a resource that already exists. Unable to
continue with install: existing resource conflict: namespace: , name:
podgroups.scheduling.incubator.k8s.io, existing_kind:
apiextensions.k8s.io/v1beta1, Kind=CustomResourceDefinition, new_kind:
apiextensions.k8s.io/v1beta1, Kind=CustomResourceDefinition
{code}
Delete the crd like this will resolve this issue:
kubectl delete crd/tfjobs.kubeflow.org && kubectl delete
crd/podgroups.scheduling.incubator.k8s.io && kubectl delete
crd/pytorchjobs.kubeflow.org
2. helm delete won't delete all resources created by submarine. Like tfjob,
pytorchjob, notebook, and ingress. We should clean up these CRD objects when
uninstallation. Otherwise, the user can only delete the whole k8s namespace to
clean up the pod.
3. We should make the CRD scope to "namespace" scope and delete them when
uninstallation.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
