kubeflow - jupyter pod got stuck after spawning sometime
I am new to the world of Kubernetes. I am trying to deploy jupyter notebook inside the cluster. I have created kubernetes cluster with reference to official docs. The notebook says it will redirect to the home page once the spawning is finished. But jupyter pod got stuck after spawning sometime.
I referred similar issue in GitHub but couldn't find the answer. The referred link is Github Link
From the comments from the issue suggested that whether the jupyter hub uses the persistent disk. I ran those commands, seems persistent disk are attached.
kubectl -n default get po,svc,deploy,pv,pvc -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE
pod/deploy-ml-pipeline-csnx4-j556r 0/1 Completed 0 30m 10.60.1.6 gke-churnprediction-default-pool-142b8f7d-d4kv <none>
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE SELECTOR
service/kubernetes ClusterIP 10.63.240.1 <none> 443/TCP 32m <none>
NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE
persistentvolume/pvc-57af1a5e-505d-11e9-9b66-42010a800130 10Gi RWO Delete Bound kubeflow/vizier-db standard 27m
persistentvolume/pvc-70874d08-505d-11e9-9b66-42010a800130 10Gi RWO Delete Bound kubeflow/minio-pv-claim standard 26m
persistentvolume/pvc-70b1712e-505d-11e9-9b66-42010a800130 10Gi RWO Delete Bound kubeflow/mysql-pv-claim standard 26m
persistentvolume/pvc-86d45ad1-505d-11e9-9b66-42010a800130 10Gi RWO Delete Bound kubeflow/claim-madhi standard 25mThis is the result of the above command which according to my knowledge persistent disk are successfully attached! I really don't know how it's internally working. So, I can't able to figure out what's the problem here. Can any explain the problem or provide the link of kubernetes architecture link? It will be helpful for me to understand the core concept behind the kubernetes.
Below is the command used to get description about pod
kubectl describe pod pod_name
to get the yaml file
kubectl get pod pod_name -o yaml
Similar Questions
1 Answer
I somewhat figured out the solution but I don't know either this is the exact solution or still there is a problem. As per the comment, there is no issue with the pod or other configuration files. I somewhat thought it might be the localhost problem. So, I tried changing the port from 8085 to 8081 and re-run the start_ui.sh script. The spawning error disappeared and it redirected me to the juypter working directory.
kubectl port-forward -n ${NAMESPACE} $(kubectl get pods -n ${NAMESPACE} --selector=service=ambassador -o jsonpath='{.items[0].metadata.name}') 8081:80And If you want to avoid all this problem, then the effective way is to the run the kubeflow in endpoints instead of localhost which eliminates all this problem. To view the dashboard at endpoints you need to setup the IAM access initially while creating a cluster.