Skip to content

Vector Store Identity

Stable vector-store identity helpers for semantic dedupe scheduling.

This module derives one stable process-local ownership key from vector-store configuration. The goal is to identify the logical store target without being sensitive to secrets or runtime-only settings.

Elasticsearch uses one provider-specific identity-field allowlist. Other providers currently fall back to hashing all non-excluded config fields after sanitization. Because of that fallback, adding a new non-excluded provider config field changes the derived scheduler key and can create a second process-local scheduler registration for the same logical store.

build_semantic_dedupe_scheduler_key_from_vector_store_config(vector_store)

Build one stable scheduler ownership key from vector-store config.

Parameters:

Name Type Description Default
vector_store dict[str, Any] | None

Native vector-store payload with provider and config sections.

required

Returns:

Type Description
str | None

str | None: Stable scheduler key, or None when the payload is unsupported.