5554aa043f
- Cache extracted triples to disk (neo4j_triples.pickle) so Neo4j can be repopulated without expensive LLM re-extraction on cold starts - Split initialization into two phases: fast vector-only (~1-2 min) and background GraphRAG, so the server serves requests while GraphRAG loads - Add GraphRAG status flags to shared_state for monitoring readiness - Update /status endpoint to expose graphrag_ready/initializing/error - Restructure main.py to use single event loop for background task support Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
22 lines
877 B
YAML
22 lines
877 B
YAML
# IMPORTANT: Neo4j data persists in a Docker named volume (hp_neo4j_data).
|
|
# This survives `docker-compose down` and `docker-compose up -d` restarts.
|
|
# However, if you delete the volume (docker volume rm hp_neo4j_data), all graph
|
|
# data (extracted triples, entities, relationships) will be lost and must be
|
|
# re-extracted via LLM calls (10-20+ minutes) or restored from the disk cache
|
|
# at index_storage/graphrag_cache/neo4j_triples.pickle.
|
|
services:
|
|
neo4j:
|
|
image: neo4j:5
|
|
container_name: hp-neo4j
|
|
restart: unless-stopped
|
|
ports:
|
|
- "7688:7687" # Bolt protocol (mapped to 7688 to avoid conflict with other Neo4j instances)
|
|
- "7475:7474" # Browser UI (mapped to 7475 to avoid conflict)
|
|
environment:
|
|
NEO4J_AUTH: neo4j/hp-graphrag-2024
|
|
NEO4J_PLUGINS: '["apoc"]'
|
|
volumes:
|
|
- hp_neo4j_data:/data
|
|
|
|
volumes:
|
|
hp_neo4j_data:
|