Skip to content

[P1] No timeouts on external HTTP service calls

Problem

Multiple services make HTTP calls to external dependencies (auth service, backend service, document repository, guardrail service) using aiohttp.ClientSession() and httpx.AsyncClient() without configuring timeouts. A hung or slow dependency can cause the calling service to hang indefinitely, eventually exhausting connection pools and rendering the service unresponsive.

Files

  • dataprep/genieai_dataprep_arangodb.py:145 — Auth token fetch: no timeout
  • dataprep/genieai_dataprep_arangodb.py:188 — Status update: no timeout
  • dataprep/genieai_dataprep_arangodb.py:216 — Ingestion log write: no timeout
  • dataprep/genieai_dataprep_arangodb.py:241 — Label fetch: no timeout
  • dataprep/genieai_dataprep_arangodb.py:318 — Guardrail check: no timeout
  • chatqna/genieai_chatqna.py:163 — Auth token fetch: no timeout
  • chatqna/genieai_chatqna.py:208 — User profile fetch: no timeout
  • chatqna/genieai_chatqna.py:841 — File metadata fetch: no timeout
  • http-service/http_service.py:9 — All HTTP calls: no timeout

Acceptance Criteria

  • Add timeout=aiohttp.ClientTimeout(total=30) (or similar) to all aiohttp.ClientSession() creation
  • Add timeout=httpx.Timeout(30.0) to all httpx.AsyncClient() creation
  • Use environment variables for configurable timeout values
  • Add circuit breaker pattern for critical dependencies (auth service)