Add a flag to enable/disable worker init container injection

[ByronHsu]

Why are these changes needed?

Currently, the init container is always injected into the worker pod with a specific command. However, in our case, ray does not exist in the $PATH so it failed. We want to have a way to disable the injection from kuberay and customize our own injection. This PR adds an env ENABLE_INIT_CONTAINER_INJECTION to switch the init container injection.

Checks

• I've made sure the tests are passing.

Running Suite: Utils Suite
==========================
Random Seed: 1683590005
Will run 4 of 4 specs

••••
Ran 4 of 4 Specs in 0.001 seconds
SUCCESS! -- 4 Passed | 0 Failed | 0 Pending | 0 Skipped

• Testing Strategy
  • Unit tests
  • Manual tests
  • This PR is not tested :(

[kevin85421]

Add this env variable in KubeRay operator Helm chart with the default value "true" and warning comments.

[kevin85421]

@ByronHsu Here is a conflict. Would you mind rebasing with the master branch?

[kevin85421]

Would you mind fixing the lint error and adding this env variable in KubeRay operator Helm chart with the default value "true" and warning comments? Thanks!

[ByronHsu]

e2e test done

1. Local build the image and push
2. Add the env and update the image to the chart

- name: ENABLE_INIT_CONTAINER_INJECTION
  value: "false"

3. Init Containers doesn't have ray health-check anymore, only containing my custom init container

Init Containers:
  init-myservice:
     <masked>

[kevin85421]

LGTM. Thank you for this contribution!

I also tested it manually.

# path: helm-chart/kuberay-operator
# Step 1: Update values.yaml
- name: ENABLE_INIT_CONTAINER_INJECTION
  value: "false"

# Step 2: Install the KubeRay operator with the new image
helm install kuberay-operator . --set image.repository=controller,image.tag=latest

# Step 3: Install a RayCluster with a init container "busybox"
helm install raycluster kuberay/ray-cluster --version 0.5.0

# Step 4: The worker Pod will only have 1 init container.

# path: helm-chart/kuberay-operator
# Step 1: Update values.yaml
- name: ENABLE_INIT_CONTAINER_INJECTION
  value: "true"

# Step 2: Install the KubeRay operator with the new image
helm install kuberay-operator . --set image.repository=controller,image.tag=latest

# Step 3: Install a RayCluster with a init container "busybox"
helm install raycluster kuberay/ray-cluster --version 0.5.0

# Step 4: The worker Pod will have 2 init containers.