Failed to start a workspace on a minikube cluster

Describe the bug

There are several issues, that user might face during starting up a workspace on the minikube cluster. Basically because of pulling Universal developer image which takes around ~15 min.

1. ImagePullBackOff error

Solution: restart the minikube cluster with the additional flag --kubernetes-version=v1.23.9

---

1. 'Starting' for longer than timeout error

Solution: Create a WorkspaceOperatorConfig:

apiVersion: controller.devfile.io/v1alpha1
kind: DevWorkspaceOperatorConfig
metadata:
  name: devworkspace-operator-config
  namespace: devworkspace-controller
config:
  workspace:
    progressTimeout: 30m

---

1. The workspace status remains "Starting" in the last 300 seconds error

Solution: Ensure that required images exist on the cluster before starting a workspace. To achieve this, we can use:

• KubernetesImagePuller https://github.com/che-incubator/kubernetes-image-puller
• Pull image locally and copy it to the cluster

docker pull ${IMAGE}
docker save "${IMAGE}" > universal-developer-image.tar
eval $(minikube docker-env) && docker load -i  universal-developer-image.tar

• Create a job to pull image:

kind: Job
apiVersion: batch/v1
metadata:
  name: prepull-universal-image
spec:
  template:
    metadata:
      name: prepull-universal-image
    spec:
      containers:
        - name: prepull-universal-image
          image: "${IMAGE}"
          imagePullPolicy: Always
      restartPolicy: OnFailure

Che version

next (development version)

Steps to reproduce

1. Deploy Eclipse Che on minikube
2. Try to start a workspace

Expected behavior

Workspace is up and running

Runtime

minikube

Screenshots

No response

Installation method

chectl/next

Environment

Linux

Eclipse Che Logs

No response

Additional context

No response

[RomanNikitenko]

Another workaround is to pull image locally and copy it to minikube

docker pull ${IMAGE}
docker save "${IMAGE}" > universal-developer-image.tar
eval $(minikube docker-env) && docker load -i  universal-developer-image.tar

I guess this part can be replaced by minikube ssh "docker pull ${IMAGE}"

I had a big image and the following commands took too long time for me:

docker save “${IMAGE}” > universal-developer-image.tar eval $(minikube docker-env) && docker load -i universal-developer-image.tar

[tolusha]

Yes it can be, but it is basically the same as https://github.com/eclipse/che/issues/21643#issuecomment-1220714353