What causes "Fork count is not implemented for Local Chroma" in chroma-core/chroma?

Point the code at a Chroma server via chromadb.HttpClient - the server tracks fork lineage and supports fork_count

chroma-core/chroma · error · NotImplementedError

Fork count is not implemented for Local Chroma

Error message

Fork count is not implemented for Local Chroma

What it means

RustBindingsAPI._fork_count (chromadb/api/rust.py:373) is a deliberate stub. Counting forks of a collection (Collection.fork_count) is a Chroma-server capability; in embedded mode with the Rust bindings there is no fork lineage to count, so the call always raises NotImplementedError.

Source

Thrown at chromadb/api/rust.py:373

    def _fork(
        self,
        collection_id: UUID,
        new_name: str,
        tenant: str = DEFAULT_TENANT,
        database: str = DEFAULT_DATABASE,
    ) -> CollectionModel:
        raise NotImplementedError(
            "Collection forking is not implemented for Local Chroma"
        )

    @override
    def _fork_count(
        self,
        collection_id: UUID,
        tenant: str = DEFAULT_TENANT,
        database: str = DEFAULT_DATABASE,
    ) -> int:
        raise NotImplementedError("Fork count is not implemented for Local Chroma")

    @override
    def _get_indexing_status(
        self,
        collection_id: UUID,
        tenant: str = DEFAULT_TENANT,
        database: str = DEFAULT_DATABASE,
    ) -> "IndexingStatus":
        raise NotImplementedError("Indexing status is not implemented for Local Chroma")

    @override
    def _search(
        self,
        collection_id: UUID,
        searches: List[Search],
        tenant: str = DEFAULT_TENANT,
        database: str = DEFAULT_DATABASE,
        read_level: ReadLevel = ReadLevel.INDEX_AND_WAL,

View on GitHub (pinned to aecdd12c8a)

Solutions

Point the code at a Chroma server via chromadb.HttpClient - the server tracks fork lineage and supports fork_count
Track fork relationships yourself in application metadata (e.g. store parent collection id in the forked collection's metadata)
Guard the call with a backend capability check and skip/return 0 when embedded

Example fix

# before (raises NotImplementedError)
count = client.get_collection('docs').fork_count()

# after - only call on server-backed clients
if type(client._server).__module__.startswith('chromadb.api.fastapi'):
    count = client.get_collection('docs').fork_count()
else:
    count = 0  # embedded mode has no fork lineage

Defensive patterns

Strategy: try-catch

Validate before calling

def fork_count_safe(collection, default: int = 0) -> int:
    try:
        return collection.fork_count()
    except NotImplementedError:
        return default  # embedded backend has no fork lineage

Try / catch

try:
    count = collection.fork_count()
except NotImplementedError as e:
    # embedded mode: no fork support
    count = 0

Prevention

Query fork lineage only from server-backed (HttpClient) deployments
Record parent/child collection relationships in collection metadata when you need lineage embedded
Centralize backend capability checks instead of scattering try/except at call sites

When it happens

Trigger: client = chromadb.RustClient(...); client.get_collection('docs').fork_count() - or the async equivalent. Any Collection.fork_count() call routed to the Rust bindings backend.

Common situations: Monitoring or lineage-audit code that worked against a Chroma server being run against an embedded Rust client; sharing utility code between server and embedded deployments.

Related errors

AI-assisted analysis of chroma-core/chroma@aecdd12c8a (2026-08-16). Data as JSON: /api/errors/7b63576167bc5c5d. Report an issue: GitHub.