[Misc] Adding script to setup ray for multi-node vllm deployments

[Edwinhr716]

FIX #8302 (link existing issues this PR will resolve)

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[ahg-g]

This is not LWS specific, it is useful for any multi-node vllm deployment

[ahg-g]

suggest to name the file multi-node-serving.sh

[Edwinhr716]

Changed

[ahg-g]

Can you please show an example? I want to make sure we support setting any vllm backend flag

[Edwinhr716]

it would be very similar to how we currently do it in the LWS repo:

command:
- sh
- -c
- "/vllm-workspace/examples/online-serving/multi-node-serving.sh leader --ray_cluster_size=$(LWS_GROUP_SIZE); 
python3 -m vllm.entrypoints.openai.api_server --port 8080 --model meta-llama/Meta-Llama-3.1-405B-Instruct --tensor-parallel-size 8 --pipeline_parallel_size 2"

[ahg-g]

Can those be changed a env vars?

[ahg-g]

@robertgshaw2-redhat do you have a minute to look into this one? It will allow us to easily launch multi-node inference workloads, see examples at https://github.com/kubernetes-sigs/lws/tree/main/docs/examples/vllm

[ahg-g]

@robertgshaw2-redhat do you have a minute to look into this one? It will allow us to easily launch multi-node inference workloads, see examples at https://github.com/kubernetes-sigs/lws/tree/main/docs/examples/vllm

Right now we have to-rebuild the container to include this script, but it would be great to have it embedded in the main vllm container.