- Remove MAX_TOKEN_SUMMARY parameter and related configurations
- Eliminate forced token-based truncation in entity/relationship descriptions
- Switch to fragment-count based summarization logic using FORCE_LLM_SUMMARY_ON_MERGE
- Update FORCE_LLM_SUMMARY_ON_MERGE default from 6 to 4 for better summarization
- Clean up documentation, environment examples, and API display code
- Preserve backward compatibility by graceful parameter removal
This change resolves issues where LLMs were forcibly truncating entity relationship
descriptions mid-sentence, leading to incomplete and potentially inaccurate knowledge
graph content. The new approach allows LLMs to generate complete descriptions while
still providing summarization when multiple fragments need to be merged.
Breaking Change: None - parameter removal is backward compatible
Fixes: Entity relationship description truncation issues
- Refactor the trigger condition for LLM-based summarization of entities and relations. Instead of relying on character length, the summary is now triggered when the number of merged description fragments exceeds a configured threshold. This provides a more robust and logical condition for consolidation.
- Introduce the `OLLAMA_NUM_CTX` environment variable to explicitly configure the context window size (`num_ctx`) for Ollama models. This decouples the model's context length from the `MAX_TOKENS` parameter, which is now specifically used to limit input for summary generation, making the configuration clearer and more flexible.
- Updated `README` files, `env.example`, and default values to reflect these changes.
- The previous implementation of the shared Redis connection pool had a critical issue where any Redis storage instance would disconnect the global shared pool upon closing. This caused `ConnectionError` exceptions for other instances still using the pool.
- This commit resolves the issue by introducing a reference counting mechanism in `RedisConnectionManager`.
- Change cleanup condition from count == 1 to count == 0 to properly
remove reused locks from cleanup list
- Fix RuntimeError: Attempting to release lock for xxxx more times than it was acquired
Move knowledge_graph_inst.has_node check inside get_storage_keyed_lock
in _merge_edges_then_upsert to ensure atomic check-then-act operations
and prevent duplicate node creation during concurrent updates.
Refactored the `KeyedUnifiedLock` to be generic and support dynamic namespaces. This decouples the locking mechanism from a specific "GraphDB" implementation, allowing it to be reused across different components and workspaces safely.
Key changes:
- `KeyedUnifiedLock` now takes a `namespace` parameter on lock acquisition.
- Renamed `_graph_db_lock_keyed` to a more generic _storage_keyed_lock`
- Replaced `get_graph_db_lock_keyed` with get_storage_keyed_lock` to support namespaces
- Add CLEANUP_THRESHOLD constant (100) to control cleanup frequency
- Modify _release_shared_raw_mp_lock to only scan when cleanup list exceeds threshold
- Modify _release_async_lock to only scan when cleanup list exceeds threshold
The document processing pipeline would crash with a TypeError when a document was submitted as raw text via the API, as the file_path attribute would be None. This change adds a check to handle the None case gracefully, preventing the crash and allowing text-based documents to be indexed correctly.
