run-llama/llama_index · error · ValueError

Failed to load index with ID {index_id}

Error message

Failed to load index with ID {index_id}

What it means

In load_indices_from_storage, when explicit index_ids are supplied, each id is looked up via storage_context.index_store.get_index_struct(index_id). A None return means no index struct with that id exists in the loaded index store, and the loop raises ValueError('Failed to load index with ID {index_id}').

Source

Thrown at llama-index-core/llama_index/core/indices/loading.py:75

    Args:
        storage_context (StorageContext): storage context containing
            docstore, index store and vector store.
        index_id (Optional[Sequence[str]]): IDs of the indices to load.
            Defaults to None, which loads all indices in the index store.
        **kwargs: Additional keyword args to pass to the index constructors.

    """
    if index_ids is None:
        logger.info("Loading all indices.")
        index_structs = storage_context.index_store.index_structs()
    else:
        logger.info(f"Loading indices with ids: {index_ids}")
        index_structs = []
        for index_id in index_ids:
            index_struct = storage_context.index_store.get_index_struct(index_id)
            if index_struct is None:
                raise ValueError(f"Failed to load index with ID {index_id}")
            index_structs.append(index_struct)

    indices = []
    for index_struct in index_structs:
        type_ = index_struct.get_type()
        index_cls = INDEX_STRUCT_TYPE_TO_INDEX_CLASS[type_]
        index = index_cls(
            index_struct=index_struct, storage_context=storage_context, **kwargs
        )
        indices.append(index)
    return indices


def load_graph_from_storage(
    storage_context: StorageContext,
    root_id: str,
    **kwargs: Any,
) -> ComposableGraph:

View on GitHub (pinned to afd0fef371)

Solutions

1. List the ids actually present and use one of them: ids = [s.index_id for s in storage_context.index_store.index_structs()].
2. Persist/track index ids at build time (index.index_id) alongside the storage dir so loaders reference the right one.
3. If no ids are known, omit index_id and use load_indices_from_storage to load everything, then choose by index type (index_struct.get_type()).

Example fix

# before
index = load_index_from_storage(storage_context, index_id="my-old-id")

# after
valid_ids = [s.index_id for s in storage_context.index_store.index_structs()]
index = load_index_from_storage(storage_context, index_id=valid_ids[0])

valid_ids = {s.index_id for s in storage_context.index_store.index_structs()}
assert index_id in valid_ids, f"{index_id} not in persisted ids: {valid_ids}"
index = load_index_from_storage(storage_context, index_id=index_id)

try:
    index = load_index_from_storage(storage_context, index_id=index_id)
except ValueError as e:
    if "Failed to load index with ID" in str(e):
        valid = [s.index_id for s in storage_context.index_store.index_structs()]
        raise RuntimeError(f"index_id stale; available: {valid}") from e
    raise

When it happens

Trigger: Calling load_index_from_storage(storage_context, index_id='...') with a stale, mistyped, or wrong-environment id; passing a node id or doc id instead of the index id; storage persisted after the index was rebuilt under a new id.

Common situations: Hardcoding an index_id that changed after a re-index; environments (dev/prod) sharing config but not storage; ids truncated or copied incorrectly; storage dir overwritten by a fresh build.

Related errors

• Expected to load a single index, but got {len(indices)} inst
• No index in storage context, check if you specified the righ
• Max iterations of {max_iterations} reached! Either something
• CitableBlock content must contain exactly one block when pro
• Eval mode {eval_mode} not supported.

AI-assisted analysis of run-llama/llama_index@afd0fef371 (2026-08-15). Data as JSON: /api/errors/4c084bddc3d9474c. Report an issue: GitHub.