docs: add multi_node examples

[tlipoca9]

Overview:

add a example to serve llm with multi node

Related Issues:

• closes GitHub issue: How to run PD disaggregating server on multiple nodes? #291

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[github-actions]

👋 Hi tlipoca9! Thank you for contributing to ai-dynamo/dynamo.

Just a reminder: The NVIDIA Test Github Validation CI runs an essential subset of the testing framework to quickly catch errors.Your PR reviewers may elect to test the changes comprehensively before approving your changes.

🚀

[ishandhanani]

Thanks for the contribution @tlipoca9!

I plan to get to this tomorrow but instead of having a sed command - what do you think of having some clear documentation around how to deploy this? Something like

Node 1

• We spin up NATS and ETCD and spin up the Frontend, Processor, and Router.
• Have a file (or just explain that you need to run Frontend.link(Processor).link(Router)

Node 2

• Here we run the workers (prefill and decode)
• Export the NATS and ETCD env vars
• Have a file (or again just explain that you need to run VllmWorker.link(PrefillWorker)

Let me create a PR for this tomorrow and tag you in case you want to take a look

[tlipoca9]

Thanks for the contribution @tlipoca9!

I plan to get to this tomorrow but instead of having a sed command - what do you think of having some clear documentation around how to deploy this? Something like

Node 1

• We spin up NATS and ETCD and spin up the Frontend, Processor, and Router.
• Have a file (or just explain that you need to run Frontend.link(Processor).link(Router)

Node 2

• Here we run the workers (prefill and decode)
• Export the NATS and ETCD env vars
• Have a file (or again just explain that you need to run VllmWorker.link(PrefillWorker)

Let me create a PR for this tomorrow and tag you in case you want to take a look

Thanks for your reply !
I think a clear documentation guides how to use dynamo to deploy PD disaggregation with mutli nodes is helpful.
Could we explicitly separate the Prefill Worker and VLLM Worker deployments on Node 2? This would clearly show the guide can used by PD disaggregation.

[llc-kc]

Does the Processor also communicates with decoding workers by NATS and ETCD? where are the related codes?