mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-20 00:11:50 +00:00
e295c2a505
PrometheusAuthMiddleware invokes user_api_key_auth(), which internally calls _read_request_body() (user_api_key_auth.py ~line 1591). That drains the ASGI `receive` channel via Starlette's Request.body(). The middleware then passed the already-drained `receive` callable straight to the downstream app, so a mounted ASGI sub-app at /metrics — the one prometheus_client.make_asgi_app() returns and that litellm/integrations/prometheus.py mounts — blocks indefinitely awaiting an http.request message that has already been consumed. In production on v1.83.7-stable with `require_auth_for_metrics_endpoint: true`, this manifests as admin-role scrapes hanging forever (TCP accepted, no access-log entry, no response) while non-admin-role scrapes return a fast 401 because auth raises before reaching _read_request_body. The fix buffers any messages auth reads off `receive` through a buffering wrapper, then hands the downstream app a replay wrapper that yields the buffered messages before falling back to the original `receive`. The 401 path is unchanged — if auth raises we respond 401 directly and never invoke the downstream app. Adds a regression test (test_downstream_asgi_app_receives_http_request_after_auth_reads_body) that installs a pure-ASGI inner app and a stub auth which mirrors the real behaviour by calling `await request.body()`. Before the middleware fix the test times out waiting on receive() inside the inner app; after the fix the inner app observes its http.request message and completes normally. The existing tests in the same module use a stub that skips the body read, which is why they never caught the bug. |
||
|---|---|---|
| .. | ||
| agent_tests | ||
| audio_tests | ||
| basic_proxy_startup_tests | ||
| batches_tests | ||
| benchmarks | ||
| code_coverage_tests | ||
| documentation_tests | ||
| enterprise | ||
| guardrails_tests | ||
| image_gen_tests | ||
| litellm | ||
| litellm-proxy-extras | ||
| litellm_core_utils | ||
| litellm_utils_tests | ||
| llm_responses_api_testing | ||
| llm_translation | ||
| load_tests | ||
| local_testing | ||
| logging_callback_tests | ||
| mcp_tests | ||
| multi_instance_e2e_tests | ||
| ocr_tests | ||
| old_proxy_tests/tests | ||
| openai_endpoints_tests | ||
| otel_tests | ||
| pass_through_tests | ||
| pass_through_unit_tests | ||
| proxy_admin_ui_tests | ||
| proxy_e2e_anthropic_messages_tests | ||
| proxy_security_tests | ||
| proxy_unit_tests | ||
| router_unit_tests | ||
| scim_tests | ||
| search_tests | ||
| spend_tracking_tests | ||
| store_model_in_db_tests | ||
| test_litellm | ||
| unified_google_tests | ||
| vector_store_tests | ||
| windows_tests | ||
| __init__.py | ||
| gettysburg.wav | ||
| large_text.py | ||
| openai_batch_completions.jsonl | ||
| README.MD | ||
| test_budget_management.py | ||
| test_callbacks_on_proxy.py | ||
| test_config.py | ||
| test_debug_warning.py | ||
| test_default_encoding_non_root.py | ||
| test_end_users.py | ||
| test_entrypoint.py | ||
| test_fallbacks.py | ||
| test_gpt5_azure_temperature_support.py | ||
| test_health.py | ||
| test_keys.py | ||
| test_litellm_proxy_responses_config.py | ||
| test_logging.conf | ||
| test_models.py | ||
| test_new_vector_store_endpoints.py | ||
| test_openai_endpoints.py | ||
| test_organizations.py | ||
| test_otel_thread_leak.py | ||
| test_passthrough_endpoints.py | ||
| test_presidio_latency.py | ||
| test_proxy_server_non_root.py | ||
| test_ratelimit.py | ||
| test_resource_cleanup.py | ||
| test_service_logger_otel.py | ||
| test_spend_logs.py | ||
| test_team.py | ||
| test_team_logging.py | ||
| test_team_members.py | ||
| test_users.py | ||
In total litellm runs 1000+ tests
[02/20/2025] Update:
To make it easier to contribute and map what behavior is tested,
we've started mapping the litellm directory in tests/test_litellm
This folder can only run mock tests.