mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-11 03:38:38 +00:00
cd7933ecef
/v1/models built a ModelInfo per listed model and each entry recomputed group metadata via the uncached Router.get_model_group_info, expanded wildcard routes with a full copy.deepcopy per matched deployment, and re-ran pattern_router.route(model_group) inside the per-deployment loop. On large deployment lists this repeated deep copying blocked the event loop for minutes and failed health checks (#33636). create_model_info_response now reads limits through the lru_cached _cached_get_model_group_info; wildcard expansion shallow-copies the deployment and its litellm_params (only litellm_params["model"] is rewritten); and _set_model_group_info drops the redundant per-deployment route re-check, since get_model_list already normalizes a matched wildcard deployment's model_name to the requested group. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|---|---|---|
| .. | ||
| pre_call_checks | ||
| test_add_retry_fallback_headers.py | ||
| test_cooldown_cache.py | ||
| test_fallback_event_handlers.py | ||
| test_health_check_allowed_fails_integration.py | ||
| test_health_state_cache.py | ||
| test_pattern_match_deployments.py | ||
| test_router_health_check_routing.py | ||
| test_router_interactions_endpoints.py | ||
| test_router_utils_common_utils.py | ||