mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-07 08:26:10 +00:00
A pytest-based load test that sends thousands of requests through litellm.acompletion() using the SleepModel CustomLLM, measuring RSS memory at baseline, during load, and after cooldown with malloc_trim. Asserts that memory growth stays under MAX_MEMORY_GROWTH_MB (default 150MB) for 2000 requests with 4KB input / 1KB output payloads. Skipped by default (RUN_MEMORY_TESTS=true to enable) since it's slow. Co-authored-by: Ishaan Jaff <ishaan-jaff@users.noreply.github.com> |
||
|---|---|---|
| .. | ||
| memory_leak_utils.py | ||
| sleep_mode.py | ||
| test_datadog_load_test.py | ||
| test_langsmith_load_test.py | ||
| test_linear_memory_growth.py | ||
| test_memory_leak_regression.py | ||
| test_memory_usage.py | ||
| test_otel_load_test.py | ||
| test_vertex_embeddings_load_test.py | ||
| test_vertex_load_tests.py | ||
| vertex_key.json | ||