vllm-project/vllm · error · ValueError

Invalid endpoint: must contain 'inproc' or 'tcp'

Error message

Invalid endpoint: must contain 'inproc' or 'tcp'

What it means

Error "Invalid endpoint: must contain 'inproc' or 'tcp'" thrown in vllm-project/vllm.

Source

Thrown at vllm/distributed/kv_events.py:525

            The endpoint with the port offset by data_parallel_rank
                or suffix appended
        """
        # Do nothing if input is None or data_parallel_rank is 0
        if not endpoint or data_parallel_rank == 0:
            return endpoint

        if "inproc" in endpoint:
            return f"{endpoint}_dp{data_parallel_rank}"
        if "tcp" in endpoint:
            if endpoint and ":" in endpoint:
                # Get everything after the last colon (the port)
                last_colon_idx = endpoint.rfind(":")
                base_addr = endpoint[:last_colon_idx]
                base_port = int(endpoint[last_colon_idx + 1 :])
                new_port = base_port + data_parallel_rank
                return f"{base_addr}:{new_port}"
            return endpoint
        raise ValueError("Invalid endpoint: must contain 'inproc' or 'tcp'")


class EventPublisherFactory:
    _registry: dict[str, Callable[..., EventPublisher]] = {
        "null": NullEventPublisher,
        "zmq": ZmqEventPublisher,
    }

    @classmethod
    def register_publisher(cls, name: str, ctor: Callable[..., EventPublisher]) -> None:
        if name in cls._registry:
            raise KeyError(f"publisher '{name}' already registered")
        cls._registry[name] = ctor

    @classmethod
    def create(
        cls, config: KVEventsConfig | None, data_parallel_rank: int = 0
    ) -> EventPublisher:

View on GitHub (pinned to c794754062)

Solutions

1. Use an endpoint containing 'inproc' or 'tcp', e.g. tcp://127.0.0.1:5557.

When it happens

Trigger: Raised at vllm/distributed/kv_events.py:525 when validation fails: Invalid endpoint: must contain 'inproc' or 'tcp'. Typically triggered by an incompatible or incomplete vLLM configuration, an unsupported platform/backend combination, or a runtime resource/dependency that is missing.

Common situations: Commonly encountered at vllm/distributed/kv_events.py:525 during vLLM startup/config validation or runtime setup when: (1) conflicting CLI flags or config fields are combined, (2) the current platform (CUDA/ROCm/CPU/XPU) or installed optional packages do not support the requested feature, or (3) a required value is absent or out of range. Resolve by correcting the configuration as described in the message, or by selecting a supported alternative.

AI-assisted analysis of vllm-project/vllm@c794754062 (2026-08-14). Data as JSON: /api/errors/b6919da3000031d7. Report an issue: GitHub.