zylon-ai/private-gpt · error · ValueError

Unsupported streaming provider: {settings.stream.broker}

Error message

Unsupported streaming provider: {settings.stream.broker}

What it means

Raised by StreamComponent.__init__ when settings.stream.broker has no entry in the _PROVIDERS registry. The component resolves the broker string to a provider callable at injection time; an unknown value means no StreamService can be constructed, so startup fails immediately with the offending broker name in the message.

Source

Thrown at private_gpt/components/streaming/stream_component.py:18

from injector import inject, singleton

from private_gpt.components.streaming.providers.stream_service import StreamService
from private_gpt.components.streaming.registry import _PROVIDERS, register_stream
from private_gpt.settings.settings import Settings

__all__ = ["StreamComponent", "register_stream"]


@singleton
class StreamComponent:
    stream: StreamService

    @inject
    def __init__(self, settings: Settings) -> None:
        provider = _PROVIDERS.get(settings.stream.broker)
        if provider is None:
            raise ValueError(
                f"Unsupported streaming provider: {settings.stream.broker}"
            )
        self.stream = provider(settings)

View on GitHub (pinned to 4a030776a3)

Solutions

1. Set stream.broker to a supported value — check the keys of _PROVIDERS in private_gpt/components/streaming/registry.py (e.g. the in-memory and redis providers)
2. Check casing/whitespace in the YAML value (values are matched exactly)
3. If you need a custom broker, register a provider function in _PROVIDERS before the component is instantiated

Example fix

# before
stream:
  broker: kafka
# after
stream:
  broker: redis   # or the in-memory provider key from registry._PROVIDERS

from private_gpt.components.streaming.registry import _PROVIDERS

if settings.stream.broker not in _PROVIDERS:
    raise ValueError(
        f"stream.broker must be one of {sorted(_PROVIDERS)}, "
        f"got {settings.stream.broker!r}"
    )

def is_supported_broker(broker: str) -> bool:
    from private_gpt.components.streaming.registry import _PROVIDERS
    return broker in _PROVIDERS

When it happens

Trigger: Setting stream.broker to anything other than a registered provider key (e.g. "kafka", "rabbitmq", "Redis" with wrong casing, or a broker removed in a version upgrade).

Common situations: Copy-pasting config from another project's messaging stack; case-sensitivity mistakes; upgrading private-gpt and a broker name was renamed or dropped from _PROVIDERS.

Related errors

• Unknown scheduler.chat.mode: {mode}
• scheduler.chat.mode={self.scheduler.chat.mode!r} requires st
• Default LLM model '{model_id}' could not be initialized: {e}
• Default model '{self._default_model_id}' not found in regist
• No model specified and no models are configured

AI-assisted analysis of zylon-ai/private-gpt@4a030776a3 (2026-08-15). Data as JSON: /api/errors/2b66c3f05410626a. Report an issue: GitHub.