Fix Installation Steps in vLLM Production Stack Tutorials

[dxu104]

I tried to reproduce the tutorials:

• [01-minimal-helm-installation](https://github.com/vllm-project/production-stack/blob/main/tutorials/01-minimal-helm-installation.md)
• [02-basic-vllm-config](https://github.com/vllm-project/production-stack/blob/main/tutorials/02-basic-vllm-config.md)

However, I encountered the following issues:

1. Helm Chart Name Issue:
   The command in the tutorial:

   helm install vllm vllm/production-stack -f tutorials/assets/values-02-basic-config.yaml

   results in an error:

   Error: INSTALLATION FAILED: chart "production-stack" matching not found in vllm index. (try 'helm repo update'): no chart name found
   

   Fix: Change the command to:

   helm install vllm vllm/vllm-stack -f tutorials/assets/values-02-basic-config.yaml

2. Persistent Volume (PV) Requirement:
   Before running helm install, we need to create a Persistent Volume (PV), or the Pod will be stuck in Pending state with the error:

   0/1 nodes are available: pod has unbound immediate PersistentVolumeClaims.
   

   Fix: Apply pv.yaml first:

   kubectl apply -f pv.yaml

   pv.yaml:

   apiVersion: v1
   kind: PersistentVolume
   metadata:
     name: vllm-pv
   spec:
     capacity:
       storage: 50Gi
     accessModes:
       - ReadWriteOnce
     persistentVolumeReclaimPolicy: Retain
     hostPath:
       path: "/mnt/data/vllm"

Would be great to update the tutorial to include these steps to help users avoid common issues!

[ApostaC]

Thanks for creating the issue!

For problem 1 (helm chart issue), you are more than welcome to create a PR for the fix. Thanks for contributing!

For problem 2, if you are running the tutorial-02, it shouldn't require creating the PV in advance if your kubernetes cluster supports dynamic volume provisioning. But if you are running tutorial-03, where the PVC needs to be matched with the PV with some specific labels, you will need to create the PV as illustrated in the tutorial.

[dxu104]

@ApostaC Thank you for your explanation.

I think maybe we installed k8s via kubeadm instead of minikube, therefore our kubernetes cluster do not supports dynamic volume provisioning.

Below PR is ready to be reviewed.
#42

[nrober734]

I think some docs on configuring the shared memory volume could also be helpful, I hit a snag here myself trying to run the tutorials: #44