This smoke test runs a complete Livepeer gateway and orchestrator on one machine using a CPU-only Docker container as a BYOC (Bring Your Own Container) pipeline. By the end, a test job has travelled from the gateway through the orchestrator into a custom container and back. Estimated time: 20 minutes. You will verify:
- The orchestrator accepts and routes BYOC jobs correctly
- Your BYOC container registers and serves requests
- The gateway receives the result from the orchestrator
Prerequisites
No GPU, Ethereum wallet, Arbitrum RPC, or on-chain registration is required.Build the CPU BYOC container
The pipeline is a CPU-only Python passthrough: frames in, frames out unchanged. This proves the job routing framework works without any model inference.Start the orchestrator
Start the gateway
Send a test job
What happened
Three processes ran on your machine and handled a complete job lifecycle:- Your BYOC container registered as a pipeline called
passthrough-cpu, exposing a health endpoint at:8000. - The orchestrator started in off-chain mode, registered the BYOC container’s capability, and listened for jobs at
:8935. - The gateway started in off-chain mode with no Ethereum wallet, connected to the community remote signer for payment operations, and pointed at your local orchestrator.
- The test request entered the gateway at
:8936, the gateway routed it to the orchestrator (with a payment ticket signed by the remote signer), the orchestrator forwarded it to the BYOC container, the container processed and returned the payload, and the result returned through the chain to the HTTP response.
Related pages
Zero to First Reward
Full production setup: install, configure, stake, activate, and claim a first LPT reward with a GPU node.
AI Earning Quickstart
Start earning from AI inference with one GPU and one warm model.
AI Inference Operations
Full aiModels.json reference and pipeline architecture for production AI workloads.
Dual Mode Configuration
Run video transcoding and AI inference from a single node.