microsoft/graphrag · error · ValueError

MetricsConfig.store '{strategy}' is not registered in the Me

Error message

MetricsConfig.store '{strategy}' is not registered in the MetricsStoreFactory. Registered strategies: {', '.join(metrics_store_factory.keys())}

What it means

This ValueError is thrown by create_metrics_store when the MetricsConfig.store value does not match any strategy registered in the MetricsStoreFactory registry. It is the catch-all branch of a match statement, so any unrecognized store type string (or enum value) reaches it. The message lists all registered strategies so the caller can correct the config.

Source

Thrown at packages/graphrag-llm/graphrag_llm/metrics/metrics_store_factory.py:81

    if config.writer:
        from graphrag_llm.metrics.metrics_writer_factory import create_metrics_writer

        metrics_writer = create_metrics_writer(config)
    init_args: dict[str, Any] = config.model_dump()

    if strategy not in metrics_store_factory:
        match strategy:
            case MetricsStoreType.Memory:
                from graphrag_llm.metrics.memory_metrics_store import MemoryMetricsStore

                register_metrics_store(
                    store_type=strategy,
                    store_initializer=MemoryMetricsStore,
                    scope="singleton",
                )
            case _:
                msg = f"MetricsConfig.store '{strategy}' is not registered in the MetricsStoreFactory. Registered strategies: {', '.join(metrics_store_factory.keys())}"
                raise ValueError(msg)

    return metrics_store_factory.create(
        strategy=strategy,
        init_args={
            **init_args,
            "id": id,
            "metrics_config": config,
            "metrics_writer": metrics_writer,
        },
    )

View on GitHub (pinned to f40e9a26ce)

Solutions

Check the error message's list of registered strategies and set MetricsConfig.store to one of them
Fix typos or outdated enum values in your YAML/JSON config or MetricsConfig instance
If you need a custom store, register it with metrics_store_factory.register(strategy=..., store_initializer=YourStore, scope=...) before calling create_completion/create_embedding
Validate config at load time (pydantic Literal/enum) so invalid store values fail fast with a clear message

Example fix

# before
config = MetricsConfig(store="mem")

# after
config = MetricsConfig(store="memory")

Defensive patterns

Strategy: validation

Validate before calling

from graphrag_llm.metrics import metrics_store_factory

strategy = metrics_config.store
if strategy not in metrics_store_factory:
    raise ValueError(f"Invalid store {strategy!r}; valid: {list(metrics_store_factory.keys())}")

Type guard

from graphrag_llm.metrics import metrics_store_factory

def is_valid_store(strategy: str) -> bool:
    return strategy in metrics_store_factory

Try / catch

try:
    store = create_metrics_store(cfg)
except ValueError as e:
    if "not registered in the MetricsStoreFactory" in str(e):
        # log config error, correct MetricsConfig.store, or re-raise
        raise
    raise

Prevention

Validate strategy fields against factory keys at config load time
Use pydantic enum/Literal types for store instead of free strings
Register custom stores once at application startup

When it happens

Trigger: Calling create_completion or create_embedding with a metrics config whose store field is misspelled, a stale enum value, or a custom type that was never registered via metrics_store_factory.register(...). Also occurs when config is loaded from YAML/JSON with an unvalidated store key.

Common situations: Typos in config files ('memmory' vs 'memory'), upgrading graphrag-llm where store enum names changed, or using a plugin/custom MetricsStore without registering it first.

Related errors

AI-assisted analysis of microsoft/graphrag@f40e9a26ce (2026-08-27). Data as JSON: /api/errors/8b26996cbf7217c9. Report an issue: GitHub.