diff --git a/.circleci/config.yml b/.circleci/config.yml index afed7853ac6..1798abe9de5 100644 --- a/.circleci/config.yml +++ b/.circleci/config.yml @@ -408,7 +408,7 @@ jobs: - run: name: Run Windows-specific test command: | - uv run --no-sync python -m pytest tests/windows_tests/ -v + uv run --no-sync python -m pytest --tb=short tests/windows_tests/ -v windows_release_wheel: executor: @@ -551,7 +551,7 @@ jobs: echo "$TEST_FILES" | circleci tests run \ --split-by=timings \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise \ --cov-report=xml \ @@ -625,7 +625,7 @@ jobs: echo "$TEST_FILES" | circleci tests run \ --split-by=timings \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise \ --cov-report=xml \ @@ -697,7 +697,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/local_testing/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit.xml \ --durations=5 \ @@ -752,7 +752,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/proxy_admin_ui_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -815,7 +815,7 @@ jobs: echo "$TEST_FILES" | circleci tests run \ --split-by=timings \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ -k 'router' \ -n 4 \ @@ -859,7 +859,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/router_unit_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -904,7 +904,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/local_testing/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit.xml \ --durations=5 \ @@ -948,7 +948,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/llm_translation/**/test_*.py" | grep -v "^tests/llm_translation/realtime/") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit.xml \ --durations=20 \ @@ -986,7 +986,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/llm_translation/realtime/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1031,7 +1031,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/agent_tests/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv -s \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1075,7 +1075,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/guardrails_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1121,7 +1121,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/unified_google_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv -s \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1176,7 +1176,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/llm_responses_api_testing/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit.xml \ --durations=5 \ @@ -1210,7 +1210,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/ocr_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1254,7 +1254,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/search_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1298,7 +1298,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/batches_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv -s \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1342,7 +1342,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/litellm_utils_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv -s \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1387,7 +1387,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/pass_through_unit_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1432,7 +1432,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/image_gen_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit.xml \ --durations=5 \ @@ -1466,7 +1466,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/logging_callback_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ -n 4 \ @@ -1511,7 +1511,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/audio_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv -s \ --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ --junitxml=test-results/junit.xml \ @@ -1531,61 +1531,6 @@ jobs: paths: - audio_coverage.xml - audio_coverage - redis_caching_unit_tests: - docker: - - *python312_image - working_directory: ~/project - - steps: - - checkout - - skip_if_unrelated_changes - - setup_google_dns - - restore_cache: - keys: - - v1-uv-cache-{{ checksum "uv.lock" }} - - install_uv - - install_rust - - run: - name: Install Dependencies - command: | - uv sync --frozen --all-groups --all-extras --python 3.12 - - save_cache: - paths: - - ~/.cache/uv - key: v1-uv-cache-{{ checksum "uv.lock" }} - # Run pytest and generate JUnit XML report - - run: - name: Run tests - command: | - mkdir -p test-results - TEST_FILES=$(printf "%s\n" \ - tests/local_testing/test_dual_cache.py \ - tests/local_testing/test_redis_batch_optimizations.py \ - tests/local_testing/test_redis_increment_with_floor.py \ - tests/local_testing/test_router_utils.py) - echo "$TEST_FILES" | circleci tests run \ - --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ - -vv -s \ - --cov=./litellm --cov=./enterprise/litellm_enterprise --cov-report=xml \ - --junitxml=test-results/junit.xml \ - --durations=5 -n 2 \ - --reruns 2 --reruns-delay 1" - no_output_timeout: 20m - - run: - name: Rename the coverage files - command: | - mv coverage.xml redis_caching_coverage.xml - mv .coverage redis_caching_coverage - - # Store test results - - store_test_results: - path: test-results - - persist_to_workspace: - root: . - paths: - - redis_caching_coverage.xml - - redis_caching_coverage installing_litellm_on_python: docker: - *python312_image @@ -1605,7 +1550,7 @@ jobs: - run: name: Run tests command: | - uv run --no-sync python -m pytest -vv tests/local_testing/test_basic_python_version.py -k "not legacy_resolver" + uv run --no-sync python -m pytest --tb=short -vv tests/local_testing/test_basic_python_version.py -k "not legacy_resolver" installing_litellm_on_python_3_13: docker: @@ -1629,7 +1574,7 @@ jobs: - run: name: Run tests command: | - uv run --no-sync python -m pytest -v tests/local_testing/test_basic_python_version.py -k "not legacy_resolver" + uv run --no-sync python -m pytest --tb=short -v tests/local_testing/test_basic_python_version.py -k "not legacy_resolver" installing_litellm_on_python_v2_migration_resolver: docker: @@ -1660,7 +1605,7 @@ jobs: - run: name: Run both migration resolvers against Postgres command: | - uv run --no-sync python -m pytest -vv \ + uv run --no-sync python -m pytest --tb=short -vv \ tests/local_testing/test_basic_python_version.py::test_litellm_proxy_server_config_no_general_settings \ tests/local_testing/test_basic_python_version.py::test_litellm_proxy_server_config_no_general_settings_legacy_resolver @@ -1829,7 +1774,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/basic_proxy_startup_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit-2.xml \ --durations=5" @@ -1926,7 +1871,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -s -v \ --junitxml=test-results/junit.xml \ -n 4 \ @@ -2013,7 +1958,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/openai_endpoints_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -s -vv \ --junitxml=test-results/junit.xml \ --durations=5" @@ -2096,7 +2041,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/otel_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit.xml \ --durations=5" @@ -2148,7 +2093,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/basic_proxy_startup_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit-2.xml \ --durations=5" @@ -2229,7 +2174,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/spend_tracking_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --junitxml=test-results/junit.xml \ --durations=5" @@ -2334,7 +2279,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/multi_instance_e2e_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --junitxml=test-results/junit.xml \ --durations=5" @@ -2406,7 +2351,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/store_model_in_db_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --junitxml=test-results/junit.xml \ --durations=5" @@ -2491,7 +2436,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/basic_proxy_startup_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv \ --junitxml=test-results/junit-2.xml \ --durations=5" @@ -2588,7 +2533,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/pass_through_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -v \ --junitxml=test-results/junit.xml \ --durations=5" @@ -2659,7 +2604,7 @@ jobs: TEST_FILES=$(circleci tests glob "tests/proxy_e2e_anthropic_messages_tests/**/test_*.py") echo "$TEST_FILES" | circleci tests run \ --verbose \ - --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest \ + --command="tr ' ' '\\n' | awk '/\\.py/ {print; next} {sub(/\\.[A-Z][^.]*$/, \"\"); gsub(/\\./, \"/\"); print \$0 \".py\"}' | xargs uv run --no-sync python -m pytest --tb=short \ -vv -s \ --junitxml=test-results/junit.xml \ --durations=5" @@ -2689,7 +2634,7 @@ jobs: - run: name: Combine Coverage command: | - uv tool run --from 'coverage[toml]==7.10.6' coverage combine realtime_translation_coverage ocr_coverage search_coverage logging_coverage audio_coverage local_testing_part1_coverage local_testing_part2_coverage pass_through_unit_tests_coverage batches_coverage guardrails_coverage redis_caching_coverage agent_coverage google_generate_content_endpoint_coverage litellm_utils_coverage router_unit_tests_coverage auth_ui_unit_tests_coverage + uv tool run --from 'coverage[toml]==7.10.6' coverage combine realtime_translation_coverage ocr_coverage search_coverage logging_coverage audio_coverage local_testing_part1_coverage local_testing_part2_coverage pass_through_unit_tests_coverage batches_coverage guardrails_coverage agent_coverage google_generate_content_endpoint_coverage litellm_utils_coverage router_unit_tests_coverage auth_ui_unit_tests_coverage uv tool run --from 'coverage[toml]==7.10.6' coverage xml - codecov/upload: file: ./coverage.xml @@ -3189,7 +3134,7 @@ jobs: name: Test provider capture and replay harness command: | mkdir -p test-results/provider-replay-harness - uv run --no-sync pytest -q --noconftest -o addopts= -o pythonpath=tests/e2e -p no:rerunfailures \ + uv run --no-sync pytest --tb=short -q --noconftest -o addopts= -o pythonpath=tests/e2e -p no:rerunfailures \ --junitxml=test-results/provider-replay-harness/junit.xml \ tests/e2e/test_provider_edge.py tests/e2e/test_fixture_bundle.py \ tests/e2e/test_fixture_canonical.py tests/e2e/test_fixture_mode.py \ @@ -3492,7 +3437,6 @@ workflows: - image_gen_testing - logging_testing - audio_testing - - redis_caching_unit_tests - upload-coverage: requires: - realtime_translation_testing @@ -3507,7 +3451,6 @@ workflows: - image_gen_testing - logging_testing - audio_testing - - redis_caching_unit_tests - langfuse_logging_unit_tests - local_testing_part1 - local_testing_part2 diff --git a/.circleci/scripts/run_integration.sh b/.circleci/scripts/run_integration.sh index 26240487c48..ba24e66ba1c 100644 --- a/.circleci/scripts/run_integration.sh +++ b/.circleci/scripts/run_integration.sh @@ -191,7 +191,7 @@ if [ "$suite" = management ] || [ "$suite" = mcp ]; then fi if [ "$suite" = providers ]; then - INTEGRATION_RUN_ID="$integration_identity" .venv/bin/python -m pytest --noconftest -o addopts= \ + INTEGRATION_RUN_ID="$integration_identity" .venv/bin/python -m pytest --tb=short --noconftest -o addopts= \ --strict-markers --strict-config -p no:pytest-retry -p no:rerunfailures --timeout=30 \ tests/e2e/test_provider_edge.py::TestReplayMode::test_content_drift_returns_the_miss_status_naming_both_keys \ tests/e2e/test_provider_edge.py::TestReplayMode::test_exhausted_key_returns_the_miss_status \ diff --git a/tests/integration/README.md b/tests/integration/README.md index c559e7545e0..2204cde11e3 100644 --- a/tests/integration/README.md +++ b/tests/integration/README.md @@ -30,7 +30,7 @@ Streaming checks send real HTTP transfer chunks, including one-byte partitions, The `messages_endpoint/` directory holds `/v1/messages` endpoint contracts: native-provider backends under `providers/` (`anthropic`, `bedrock`, `gemini`) and the translation bridges (`responses_bridge`, `chat_bridge`) at the top level. It runs in the providers shard; `run.py` selects test files recursively under each scheduled directory -The sdk shard exercises the SDK's own HTTP clients against local protocol peers with no gateway in the path, so a case here fails only when the client library or its wire behavior changes. The HTTP/2 case runs a hypercorn TLS peer offering h2 and http/1.1 over ALPN, drives the sync and async httpx handlers at it with `LITELLM_HTTP2` off and on, and asserts the version both the client and the peer observed on the wire. Put a test here only when it needs no proxy, database or Redis; a case that reaches the gateway belongs in one of the other shards +The sdk shard exercises the SDK's own HTTP clients against local protocol peers with no gateway in the path, so a case here fails only when the client library or its wire behavior changes. The HTTP/2 case runs a hypercorn TLS peer offering h2 and http/1.1 over ALPN, drives the sync and async httpx handlers at it with `LITELLM_HTTP2` off and on, and asserts the version both the client and the peer observed on the wire. Put a test here only when it needs no proxy or database. CircleCI starts a local Redis for this shard like the others, so SDK-side caching cases that need a real Redis server belong here too; a case that reaches the gateway belongs in one of the other shards The extensions shard uses the built-in generic callback and guardrail transports. It checks callback correlation and credential exclusion, guardrail rewriting and denial, retained OpenAI consumers and A2A wire versions. CircleCI runs it on parallel nodes, and each node starts its own database, Redis, upstream and proxy and runs its share of the group's files serially, split by recorded timings with `circleci tests split`. Tests keep the isolation of a serial run; they still must not assume a particular set of sibling files. `run.py --list` prints a group's files and `run.py ...` runs a subset of them diff --git a/tests/local_testing/test_redis_increment_with_floor.py b/tests/integration/routing/test_redis_increment_with_floor.py similarity index 65% rename from tests/local_testing/test_redis_increment_with_floor.py rename to tests/integration/routing/test_redis_increment_with_floor.py index e358d5f31e0..d5535d30715 100644 --- a/tests/local_testing/test_redis_increment_with_floor.py +++ b/tests/integration/routing/test_redis_increment_with_floor.py @@ -1,15 +1,9 @@ -"""Least-busy routing keeps its in-flight counters in Redis, and the clamp at zero plus the -create-once TTL both live inside a Lua script. Nothing but a real Redis runs that script, so -these are the only tests that fail when the script itself is wrong.""" - import os import uuid +from collections.abc import Iterator from typing import Final import pytest -from dotenv import load_dotenv - -load_dotenv() from litellm.caching.redis_cache import RedisCache @@ -17,14 +11,14 @@ TTL: Final = 600 @pytest.fixture -def counter(): - cache: Final = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - key: Final = f"lit7039-{uuid.uuid4()}" +def counter() -> Iterator[tuple[RedisCache, str, str]]: + cache: Final = RedisCache(host=os.environ["REDIS_HOST"], port=int(os.environ["REDIS_PORT"])) + key: Final = f"increment-with-floor-{uuid.uuid4()}" yield cache, key, cache.check_and_fix_namespace(key=key) cache.delete_cache(key) -def test_a_counter_adds_every_increment_and_reads_back_what_it_holds(counter): +def test_a_counter_adds_every_increment_and_reads_back_what_it_holds(counter: tuple[RedisCache, str, str]) -> None: cache, key, _ = counter assert cache.increment_with_floor(key, 3, TTL) == 3 @@ -32,10 +26,7 @@ def test_a_counter_adds_every_increment_and_reads_back_what_it_holds(counter): assert cache.batch_get_counts([key]) == (5,) -def test_a_decrement_past_zero_leaves_the_counter_at_zero(counter): - """A worker whose counter expired mid-request decrements a key that is no longer there. - Without the clamp that deployment reads negative, and least-busy pins every later request - on it until the count climbs back to zero.""" +def test_a_decrement_past_zero_leaves_the_counter_at_zero(counter: tuple[RedisCache, str, str]) -> None: cache, key, _ = counter assert cache.increment_with_floor(key, 1, TTL) == 1 @@ -43,9 +34,7 @@ def test_a_decrement_past_zero_leaves_the_counter_at_zero(counter): assert cache.batch_get_counts([key]) == (0,) -def test_traffic_never_pushes_a_counters_expiry_back_out(counter): - """The TTL is what releases a count whose worker died mid-request. Rewriting it on every - touch would keep that stuck count alive for as long as the group takes traffic.""" +def test_traffic_never_pushes_a_counters_expiry_back_out(counter: tuple[RedisCache, str, str]) -> None: cache, key, namespaced_key = counter cache.increment_with_floor(key, 1, TTL) @@ -57,7 +46,7 @@ def test_traffic_never_pushes_a_counters_expiry_back_out(counter): assert cache.redis_client.ttl(namespaced_key) <= 30 -def test_clamping_to_zero_keeps_the_expiry_it_already_had(counter): +def test_clamping_to_zero_keeps_the_expiry_it_already_had(counter: tuple[RedisCache, str, str]) -> None: cache, key, namespaced_key = counter cache.increment_with_floor(key, 1, TTL) @@ -68,7 +57,7 @@ def test_clamping_to_zero_keeps_the_expiry_it_already_had(counter): @pytest.mark.asyncio -async def test_the_async_counter_behaves_the_same_way(counter): +async def test_the_async_counter_behaves_the_same_way(counter: tuple[RedisCache, str, str]) -> None: cache, key, namespaced_key = counter assert await cache.async_increment_with_floor(key, 2, TTL) == 2 diff --git a/tests/integration/run.py b/tests/integration/run.py index 19bce35f542..30c1352f048 100644 --- a/tests/integration/run.py +++ b/tests/integration/run.py @@ -72,6 +72,7 @@ def main() -> int: "no:rerunfailures", "--timeout=90", "--durations=15", + "--tb=short", f"--hypothesis-seed={options.seed}", f"--integration-order-seed={options.order_seed}", f"--junitxml={output / 'junit.xml'}", diff --git a/tests/integration/sdk/test_dual_cache_redis.py b/tests/integration/sdk/test_dual_cache_redis.py new file mode 100644 index 00000000000..ddd01480d36 --- /dev/null +++ b/tests/integration/sdk/test_dual_cache_redis.py @@ -0,0 +1,94 @@ +import asyncio +import os +import uuid +from typing import Final +from unittest.mock import patch + +import pytest + +from litellm.caching.dual_cache import DualCache +from litellm.caching.in_memory_cache import InMemoryCache +from litellm.caching.redis_cache import RedisCache + + +def _redis_cache() -> RedisCache: + return RedisCache(host=os.environ["REDIS_HOST"], port=int(os.environ["REDIS_PORT"])) + + +@pytest.mark.asyncio +async def test_a_value_only_in_redis_is_read_once_from_redis_then_from_memory() -> None: + redis_cache: Final = _redis_cache() + dual_cache: Final = DualCache(in_memory_cache=InMemoryCache(), redis_cache=redis_cache) + sync_key: Final = f"redis-only-sync-{uuid.uuid4()}" + async_key: Final = f"redis-only-async-{uuid.uuid4()}" + redis_cache.set_cache(sync_key, {"v": "sync"}) + await redis_cache.async_set_cache(async_key, {"v": "async"}) + + assert dual_cache.get_cache(sync_key) == {"v": "sync"} + assert await dual_cache.async_get_cache(async_key) == {"v": "async"} + + with ( + patch.object(redis_cache, "get_cache") as sync_redis_read, + patch.object(redis_cache, "async_get_cache") as async_redis_read, + ): + assert dual_cache.get_cache(sync_key) == {"v": "sync"} + assert await dual_cache.async_get_cache(async_key) == {"v": "async"} + sync_redis_read.assert_not_called() + async_redis_read.assert_not_called() + + +@pytest.mark.asyncio +async def test_a_deleted_key_is_gone_from_both_memory_and_redis() -> None: + redis_cache: Final = _redis_cache() + dual_cache: Final = DualCache(in_memory_cache=InMemoryCache(), redis_cache=redis_cache) + sync_key: Final = f"deleted-sync-{uuid.uuid4()}" + async_key: Final = f"deleted-async-{uuid.uuid4()}" + dual_cache.set_cache(sync_key, {"v": "sync"}) + await dual_cache.async_set_cache(async_key, {"v": "async"}) + + dual_cache.delete_cache(sync_key) + await dual_cache.async_delete_cache(async_key) + + assert dual_cache.get_cache(sync_key) is None + assert await dual_cache.async_get_cache(async_key) is None + assert redis_cache.get_cache(sync_key) is None + assert await redis_cache.async_get_cache(async_key) is None + + +@pytest.mark.asyncio +async def test_a_batch_read_without_an_in_memory_cache_reads_redis() -> None: + redis_cache: Final = _redis_cache() + dual_cache: Final = DualCache(in_memory_cache=None, redis_cache=redis_cache) + key: Final = f"no-memory-{uuid.uuid4()}" + await redis_cache.async_set_cache(key, {"v": "from-redis"}) + + assert await dual_cache.async_batch_get_cache([key]) == [{"v": "from-redis"}] + + +@pytest.mark.asyncio +async def test_sync_and_async_batch_reads_share_one_redis_without_sync_reads_going_async() -> None: + redis_cache: Final = _redis_cache() + dual_cache: Final = DualCache(redis_cache=redis_cache) + run_id: Final = uuid.uuid4().hex + sync_keys: Final = [f"sync_{run_id}_{index}" for index in range(5)] + async_keys: Final = [f"async_{run_id}_{index}" for index in range(5)] + in_loop_keys: Final = [f"in_loop_{run_id}_{index}" for index in range(3)] + survivor_key: Final = f"survivor_{run_id}" + expected: Final = {key: {"key": key} for key in [*sync_keys, *async_keys, *in_loop_keys, survivor_key]} + await asyncio.gather(*(redis_cache.async_set_cache(key, value, ttl=60) for key, value in expected.items())) + + concurrent_results: Final = await asyncio.gather( + *(asyncio.to_thread(dual_cache.batch_get_cache, keys=[key]) for key in sync_keys), + *(dual_cache.async_batch_get_cache(keys=[key]) for key in async_keys), + ) + assert list(concurrent_results) == [[expected[key]] for key in [*sync_keys, *async_keys]] + + with patch.object( + redis_cache, + "async_batch_get_cache", + side_effect=AssertionError("sync batch reads must not call async Redis"), + ): + in_loop_results: Final = [dual_cache.batch_get_cache(keys=[key]) for key in in_loop_keys] + + assert in_loop_results == [[expected[key]] for key in in_loop_keys] + assert await dual_cache.async_batch_get_cache(keys=[survivor_key]) == [expected[survivor_key]] diff --git a/tests/local_testing/test_dual_cache.py b/tests/local_testing/test_dual_cache.py deleted file mode 100644 index 43b10a9557a..00000000000 --- a/tests/local_testing/test_dual_cache.py +++ /dev/null @@ -1,274 +0,0 @@ -import os -import time -import traceback -from litellm._uuid import uuid - -from dotenv import load_dotenv - -load_dotenv() - -import asyncio -import hashlib -import random - -import pytest - -import litellm -from litellm import aembedding, completion, embedding -from litellm.caching.caching import Cache - -from unittest.mock import AsyncMock, patch, MagicMock, call -import datetime -from datetime import timedelta -from litellm.caching import * - - -@pytest.mark.parametrize("is_async", [True, False]) -@pytest.mark.asyncio -async def test_dual_cache_get_set(is_async): - """Test that DualCache reads from in-memory cache first for both sync and async operations""" - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(in_memory_cache=in_memory, redis_cache=redis_cache) - - # Test basic set/get - test_key = f"test_key_{str(uuid.uuid4())}" - test_value = {"test": "value"} - - if is_async: - await dual_cache.async_set_cache(test_key, test_value) - mock_method = "async_get_cache" - else: - dual_cache.set_cache(test_key, test_value) - mock_method = "get_cache" - - # Mock Redis get to ensure we're not calling it - # this should only read in memory since we just set test_key - with patch.object(redis_cache, mock_method) as mock_redis_get: - if is_async: - result = await dual_cache.async_get_cache(test_key) - else: - result = dual_cache.get_cache(test_key) - - assert result == test_value - mock_redis_get.assert_not_called() # Verify Redis wasn't accessed - - -@pytest.mark.parametrize("is_async", [True, False]) -@pytest.mark.asyncio -async def test_dual_cache_local_only(is_async): - """Test that when local_only=True, only in-memory cache is used""" - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(in_memory_cache=in_memory, redis_cache=redis_cache) - - test_key = f"test_key_{str(uuid.uuid4())}" - test_value = {"test": "value"} - - # Mock Redis methods to ensure they're not called - redis_set_method = "async_set_cache" if is_async else "set_cache" - redis_get_method = "async_get_cache" if is_async else "get_cache" - - with ( - patch.object(redis_cache, redis_set_method) as mock_redis_set, - patch.object(redis_cache, redis_get_method) as mock_redis_get, - ): - - # Set value with local_only=True - if is_async: - await dual_cache.async_set_cache(test_key, test_value, local_only=True) - result = await dual_cache.async_get_cache(test_key, local_only=True) - else: - dual_cache.set_cache(test_key, test_value, local_only=True) - result = dual_cache.get_cache(test_key, local_only=True) - - assert result == test_value - mock_redis_set.assert_not_called() # Verify Redis set wasn't called - mock_redis_get.assert_not_called() # Verify Redis get wasn't called - - -@pytest.mark.parametrize("is_async", [True, False]) -@pytest.mark.asyncio -async def test_dual_cache_value_not_in_memory(is_async): - """Test that DualCache falls back to Redis when value isn't in memory, - and subsequent requests use in-memory cache""" - - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(in_memory_cache=in_memory, redis_cache=redis_cache) - - test_key = f"test_key_{str(uuid.uuid4())}" - test_value = {"test": "value"} - - # First, set value only in Redis - if is_async: - await redis_cache.async_set_cache(test_key, test_value) - else: - redis_cache.set_cache(test_key, test_value) - - # First request - should fall back to Redis and populate in-memory - if is_async: - result = await dual_cache.async_get_cache(test_key) - else: - result = dual_cache.get_cache(test_key) - - assert result == test_value - - # Second request - should now use in-memory cache - with patch.object( - redis_cache, "async_get_cache" if is_async else "get_cache" - ) as mock_redis_get: - if is_async: - result = await dual_cache.async_get_cache(test_key) - else: - result = dual_cache.get_cache(test_key) - - assert result == test_value - mock_redis_get.assert_not_called() # Verify Redis wasn't accessed second time - - -@pytest.mark.parametrize("is_async", [True, False]) -@pytest.mark.asyncio -async def test_dual_cache_batch_operations(is_async): - """Test batch get/set operations use in-memory cache correctly""" - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(in_memory_cache=in_memory, redis_cache=redis_cache) - - test_keys = [f"test_key_{str(uuid.uuid4())}" for _ in range(3)] - test_values = [{"test": f"value_{i}"} for i in range(3)] - cache_list = list(zip(test_keys, test_values)) - - # Set values - if is_async: - await dual_cache.async_set_cache_pipeline(cache_list) - else: - for key, value in cache_list: - dual_cache.set_cache(key, value) - - # Verify in-memory cache is used for subsequent reads - with patch.object( - redis_cache, "async_batch_get_cache" if is_async else "batch_get_cache" - ) as mock_redis_get: - if is_async: - results = await dual_cache.async_batch_get_cache(test_keys) - else: - results = dual_cache.batch_get_cache(test_keys, parent_otel_span=None) - - assert results == test_values - mock_redis_get.assert_not_called() - - -@pytest.mark.parametrize("is_async", [True, False]) -@pytest.mark.asyncio -async def test_dual_cache_increment(is_async): - """Test increment operations only use in memory when local_only=True""" - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(in_memory_cache=in_memory, redis_cache=redis_cache) - - test_key = f"counter_{str(uuid.uuid4())}" - increment_value = 1 - - # increment should use in-memory cache - with patch.object( - redis_cache, "async_increment" if is_async else "increment_cache" - ) as mock_redis_increment: - if is_async: - result = await dual_cache.async_increment_cache( - test_key, - increment_value, - local_only=True, - parent_otel_span=None, - ) - else: - result = dual_cache.increment_cache( - test_key, increment_value, local_only=True - ) - - assert result == increment_value - mock_redis_increment.assert_not_called() - - -@pytest.mark.asyncio -async def test_dual_cache_sadd(): - """Test set add operations use in-memory cache for reads""" - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(in_memory_cache=in_memory, redis_cache=redis_cache) - - test_key = f"set_{str(uuid.uuid4())}" - test_values = ["value1", "value2", "value3"] - - # Add values to set - await dual_cache.async_set_cache_sadd(test_key, test_values) - - # Verify in-memory cache is used for subsequent operations - with patch.object(redis_cache, "async_get_cache") as mock_redis_get: - result = await dual_cache.async_get_cache(test_key) - assert set(result) == set(test_values) - mock_redis_get.assert_not_called() - - -@pytest.mark.parametrize("is_async", [True, False]) -@pytest.mark.asyncio -async def test_dual_cache_delete(is_async): - """Test delete operations remove from both caches""" - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(in_memory_cache=in_memory, redis_cache=redis_cache) - - test_key = f"test_key_{str(uuid.uuid4())}" - test_value = {"test": "value"} - - # Set value - if is_async: - await dual_cache.async_set_cache(test_key, test_value) - else: - dual_cache.set_cache(test_key, test_value) - - # Delete value - if is_async: - await dual_cache.async_delete_cache(test_key) - else: - dual_cache.delete_cache(test_key) - - # Verify value is deleted from both caches - if is_async: - result = await dual_cache.async_get_cache(test_key) - else: - result = dual_cache.get_cache(test_key) - - assert result is None - - -@pytest.mark.asyncio -async def test_dual_cache_concurrent_sync_and_async_redis_reads(): - """Sync and async batch reads share one Redis backend in one process, and sync reads never open an async connection""" - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache(redis_cache=redis_cache) - - run_id = str(uuid.uuid4()) - sync_keys = [f"sync_{run_id}_{index}" for index in range(5)] - async_keys = [f"async_{run_id}_{index}" for index in range(5)] - in_loop_keys = [f"in_loop_{run_id}_{index}" for index in range(3)] - survivor_key = f"survivor_{run_id}" - expected = {key: {"key": key} for key in [*sync_keys, *async_keys, *in_loop_keys, survivor_key]} - for key, value in expected.items(): - await redis_cache.async_set_cache(key, value, ttl=60) - - concurrent_results = await asyncio.gather( - *(asyncio.to_thread(dual_cache.batch_get_cache, keys=[key]) for key in sync_keys), - *(dual_cache.async_batch_get_cache(keys=[key]) for key in async_keys), - ) - assert list(concurrent_results) == [[expected[key]] for key in [*sync_keys, *async_keys]] - - with patch.object( - redis_cache, - "async_batch_get_cache", - side_effect=AssertionError("sync batch reads must not call async Redis"), - ): - in_loop_results = [dual_cache.batch_get_cache(keys=[key]) for key in in_loop_keys] - - assert in_loop_results == [[expected[key]] for key in in_loop_keys] - assert await dual_cache.async_batch_get_cache(keys=[survivor_key]) == [expected[survivor_key]] diff --git a/tests/local_testing/test_redis_batch_optimizations.py b/tests/local_testing/test_redis_batch_optimizations.py deleted file mode 100644 index d49939cff1a..00000000000 --- a/tests/local_testing/test_redis_batch_optimizations.py +++ /dev/null @@ -1,123 +0,0 @@ -""" -Tests for Redis batch caching optimizations (commit 3f52e8c) - -Verifies: - -1. Batch cache size increased from 100 → 1000 (minimum 1k) -2. Repeated Redis queries for cache misses are throttled -""" - -import os -import time -from unittest.mock import AsyncMock, patch - -import pytest -from dotenv import load_dotenv - -load_dotenv() - -import uuid -from litellm.caching.dual_cache import DualCache -from litellm.caching.in_memory_cache import InMemoryCache -from litellm.caching.redis_cache import RedisCache -from litellm.constants import DEFAULT_MAX_REDIS_BATCH_CACHE_SIZE - - -@pytest.fixture -def cache_setup(): - """Create cache instances for testing""" - in_memory = InMemoryCache() - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - dual_cache = DualCache( - in_memory_cache=in_memory, - redis_cache=redis_cache, - default_max_redis_batch_cache_size=DEFAULT_MAX_REDIS_BATCH_CACHE_SIZE, - ) - return dual_cache, in_memory, redis_cache - - -@pytest.mark.asyncio -async def test_batch_cache_size_is_1000_minimum(cache_setup): - """Verify batch cache size is set to 1000 (never below 1k)""" - dual_cache, _, _ = cache_setup - - # Critical: batch cache size must be at least DEFAULT_MAX_REDIS_BATCH_CACHE_SIZE - assert ( - dual_cache.last_redis_batch_access_time.max_size - >= DEFAULT_MAX_REDIS_BATCH_CACHE_SIZE - ) - - -@pytest.mark.asyncio -async def test_throttling_prevents_duplicate_redis_calls(cache_setup): - """Test throttling prevents repeated Redis queries for cache misses""" - dual_cache, _, redis_cache = cache_setup - - test_keys = [f"miss_{str(uuid.uuid4())}" for _ in range(3)] - - # Set short expiry for testing - dual_cache.redis_batch_cache_expiry = 0.1 # 100ms - - with patch.object( - redis_cache, "async_batch_get_cache", new_callable=AsyncMock - ) as mock_redis: - mock_redis.return_value = {key: None for key in test_keys} - - # First call hits Redis (no throttle data exists) - await dual_cache.async_batch_get_cache(test_keys) - assert mock_redis.call_count == 1 - - # Second call immediately - throttled (within expiry window) - await dual_cache.async_batch_get_cache(test_keys) - assert mock_redis.call_count == 1 - - # Verify all keys tracked in throttle cache - for key in test_keys: - assert key in dual_cache.last_redis_batch_access_time - - # Wait for expiry time to pass - time.sleep(0.15) - - # Third call after expiry - call_count increases to 2 - await dual_cache.async_batch_get_cache(test_keys) - assert mock_redis.call_count == 2 - - -@pytest.mark.asyncio -async def test_basic_functionality_not_broken(cache_setup): - """Ensure basic cache functionality still works after optimizations""" - dual_cache, _, _ = cache_setup - - # Test basic set/get works - test_key = f"functional_test_{str(uuid.uuid4())}" - test_value = {"test": "data"} - - await dual_cache.async_set_cache(test_key, test_value) - result = await dual_cache.async_get_cache(test_key) - - assert result == test_value - - -@pytest.mark.asyncio -async def test_batch_get_with_no_in_memory_cache(): - """Test that batch get works when in_memory_cache is None""" - redis_cache = RedisCache(host=os.getenv("REDIS_HOST"), port=os.getenv("REDIS_PORT")) - - # Create DualCache with no in-memory cache - dual_cache = DualCache( - in_memory_cache=None, # This is the edge case we're testing - redis_cache=redis_cache, - ) - - # Set some test data directly in Redis - test_key = f"no_memory_test_{str(uuid.uuid4())}" - test_value = {"test": "data_without_memory_cache"} - - await redis_cache.async_set_cache(test_key, test_value) - - # Should not crash when fetching from Redis without in-memory cache - result = await dual_cache.async_batch_get_cache([test_key]) - - assert result is not None - assert len(result) == 1 - assert result[0] == test_value diff --git a/tests/local_testing/test_router_utils.py b/tests/local_testing/test_router_utils.py index 635bda55144..aa617b09731 100644 --- a/tests/local_testing/test_router_utils.py +++ b/tests/local_testing/test_router_utils.py @@ -18,73 +18,6 @@ from unittest.mock import patch, MagicMock, AsyncMock load_dotenv() -def test_returned_settings(): - # this tests if the router raises an exception when invalid params are set - # in this test both deployments have bad keys - Keep this test. It validates if the router raises the most recent exception - litellm.set_verbose = True - import openai - - try: - print("testing if router raises an exception") - model_list = [ - { - "model_name": "gpt-3.5-turbo", # openai model name - "litellm_params": { # params for litellm completion/embedding call - "model": "azure/gpt-4.1-mini", - "api_key": "bad-key", - "api_version": os.getenv("AZURE_API_VERSION"), - "api_base": os.getenv("AZURE_AI_API_BASE"), - }, - "tpm": 240000, - "rpm": 1800, - }, - { - "model_name": "gpt-3.5-turbo", # openai model name - "litellm_params": { # - "model": "gpt-3.5-turbo", - "api_key": "bad-key", - }, - "tpm": 240000, - "rpm": 1800, - }, - ] - router = Router( - model_list=model_list, - redis_host=os.getenv("REDIS_HOST"), - redis_password=os.getenv("REDIS_PASSWORD"), - redis_port=int(os.getenv("REDIS_PORT")), - routing_strategy="latency-based-routing", - routing_strategy_args={"ttl": 10}, - set_verbose=False, - num_retries=3, - retry_after=5, - allowed_fails=1, - cooldown_time=30, - ) # type: ignore - - settings = router.get_settings() - print(settings) - - """ - routing_strategy: "simple-shuffle" - routing_strategy_args: {"ttl": 10} # Average the last 10 calls to compute avg latency per model - allowed_fails: 1 - num_retries: 3 - retry_after: 5 # seconds to wait before retrying a failed request - cooldown_time: 30 # seconds to cooldown a deployment after failure - """ - assert settings["routing_strategy"] == "latency-based-routing" - assert settings["routing_strategy_args"]["ttl"] == 10 - assert settings["allowed_fails"] == 1 - assert settings["num_retries"] == 3 - assert settings["retry_after"] == 5 - assert settings["cooldown_time"] == 30 - - except Exception: - print(traceback.format_exc()) - pytest.fail("An error occurred - " + traceback.format_exc()) - - from litellm.types.utils import CallTypes diff --git a/tests/unit/caching/test_dual_cache.py b/tests/unit/caching/test_dual_cache.py index 521fda31b58..46600e0bf60 100644 --- a/tests/unit/caching/test_dual_cache.py +++ b/tests/unit/caching/test_dual_cache.py @@ -925,3 +925,100 @@ async def test_shared_batch_read_keeps_a_caches_own_tier_failure_to_itself_like_ assert shared == separate == [None, None, [3]] assert redis.async_batch_get_cache.await_args_list[0].args[0] == ["b1", "c1"] + + +def _write_through_dual_cache() -> tuple[DualCache, MagicMock]: + redis_cache: Final = MagicMock(spec=RedisCache) + return DualCache(in_memory_cache=InMemoryCache(), redis_cache=redis_cache), redis_cache + + +@pytest.mark.asyncio +async def test_a_written_value_is_read_back_from_memory_without_a_redis_read(): + dual_cache, redis_cache = _write_through_dual_cache() + + dual_cache.set_cache("sync-key", {"v": 1}) + await dual_cache.async_set_cache("async-key", {"v": 2}) + + assert dual_cache.get_cache("sync-key") == {"v": 1} + assert await dual_cache.async_get_cache("async-key") == {"v": 2} + redis_cache.set_cache.assert_called_once() + redis_cache.async_set_cache.assert_awaited_once() + redis_cache.get_cache.assert_not_called() + redis_cache.async_get_cache.assert_not_called() + + +@pytest.mark.asyncio +async def test_local_only_reads_and_writes_never_reach_redis(): + dual_cache, redis_cache = _write_through_dual_cache() + + dual_cache.set_cache("sync-key", "sync", local_only=True) + await dual_cache.async_set_cache("async-key", "async", local_only=True) + + assert dual_cache.get_cache("sync-key", local_only=True) == "sync" + assert await dual_cache.async_get_cache("async-key", local_only=True) == "async" + assert dual_cache.get_cache("missing", local_only=True) is None + assert await dual_cache.async_get_cache("missing", local_only=True) is None + redis_cache.set_cache.assert_not_called() + redis_cache.async_set_cache.assert_not_called() + redis_cache.get_cache.assert_not_called() + redis_cache.async_get_cache.assert_not_called() + + +@pytest.mark.asyncio +async def test_batch_reads_of_written_keys_are_served_from_memory(): + dual_cache, redis_cache = _write_through_dual_cache() + entries: Final = (("a", {"v": "a"}), ("b", {"v": "b"}), ("c", {"v": "c"})) + + await dual_cache.async_set_cache_pipeline(entries) + dual_cache.set_cache("d", {"v": "d"}) + + assert await dual_cache.async_batch_get_cache(["a", "b", "c"]) == [{"v": "a"}, {"v": "b"}, {"v": "c"}] + assert dual_cache.batch_get_cache(["d"], parent_otel_span=None) == [{"v": "d"}] + redis_cache.async_set_cache_pipeline.assert_awaited_once() + redis_cache.async_batch_get_cache.assert_not_called() + redis_cache.batch_get_cache.assert_not_called() + + +@pytest.mark.asyncio +async def test_local_only_increments_count_in_memory_without_touching_redis(): + dual_cache, redis_cache = _write_through_dual_cache() + + assert dual_cache.increment_cache("sync-counter", 2, local_only=True) == 2 + assert dual_cache.increment_cache("sync-counter", 3, local_only=True) == 5 + assert await dual_cache.async_increment_cache("async-counter", 4, local_only=True) == 4 + redis_cache.increment_cache.assert_not_called() + redis_cache.async_increment.assert_not_called() + + +@pytest.mark.asyncio +async def test_set_members_added_through_the_dual_cache_are_read_from_memory(): + dual_cache, redis_cache = _write_through_dual_cache() + + await dual_cache.async_set_cache_sadd("members", ["value1", "value2", "value3"]) + + assert set(await dual_cache.async_get_cache("members")) == {"value1", "value2", "value3"} + redis_cache.async_set_cache_sadd.assert_awaited_once() + redis_cache.async_get_cache.assert_not_called() + + +def test_the_batch_read_throttle_tracks_at_least_the_default_number_of_keys(): + assert DualCache().last_redis_batch_access_time.max_size >= DEFAULT_MAX_REDIS_BATCH_CACHE_SIZE + + +@pytest.mark.asyncio +async def test_async_batch_reads_of_missing_keys_hit_redis_once_per_expiry_window(): + redis_cache: Final = MagicMock(spec=RedisCache) + keys: Final = ["miss-a", "miss-b", "miss-c"] + redis_cache.async_batch_get_cache = AsyncMock(return_value=dict.fromkeys(keys)) + dual_cache: Final = DualCache( + in_memory_cache=InMemoryCache(), redis_cache=redis_cache, default_redis_batch_cache_expiry=60 + ) + + await dual_cache.async_batch_get_cache(keys) + await dual_cache.async_batch_get_cache(keys) + assert redis_cache.async_batch_get_cache.await_count == 1 + assert all(key in dual_cache.last_redis_batch_access_time for key in keys) + + dual_cache.last_redis_batch_access_time.update({key: time.time() - 61 for key in keys}) + await dual_cache.async_batch_get_cache(keys) + assert redis_cache.async_batch_get_cache.await_count == 2 diff --git a/tests/unit/test_router_get_settings.py b/tests/unit/test_router_get_settings.py new file mode 100644 index 00000000000..a4675715490 --- /dev/null +++ b/tests/unit/test_router_get_settings.py @@ -0,0 +1,26 @@ +from typing import Final + +from litellm import Router + + +def test_get_settings_returns_the_routing_and_retry_settings_the_router_was_built_with(): + router: Final = Router( + model_list=[ + {"model_name": "gpt-4.1-mini", "litellm_params": {"model": "openai/gpt-4.1-mini", "api_key": "fake-key"}} + ], + routing_strategy="latency-based-routing", + routing_strategy_args={"ttl": 10}, + num_retries=3, + retry_after=5, + allowed_fails=1, + cooldown_time=30, + ) + + settings: Final = router.get_settings() + + assert settings["routing_strategy"] == "latency-based-routing" + assert settings["routing_strategy_args"]["ttl"] == 10 + assert settings["allowed_fails"] == 1 + assert settings["num_retries"] == 3 + assert settings["retry_after"] == 5 + assert settings["cooldown_time"] == 30