Yassin Kortam 40b8300ac2 fix(spend): bound each spend-log write statement by row count as well as bytes (#37758) ... The Prisma query engine is a separate process whose resident memory grows with what it is asked to hold and glibc never returns it, so a pod's memory floor ratchets up to its worst statement and stays there for the life of the worker. #34956 bounded a spend-log flush by payload bytes, which caps that floor when prompts are stored and does nothing when they are not: rows carrying only attribution metadata run about 1.2 KB, so a 1000-row statement is roughly 1.2 MB, the 2 MB byte budget never binds, and every statement stays at 1000 rows forever. The engine charges per row as well as per byte. Measured on a container running the same engine build (5.4.2) against real Postgres, with rows shaped like a store_prompts_in_spend_logs=false deployment, writing the same 200,000 rows: rows/statement engine RSS still resident after the flush 1000 179 MB 500 91 MB 250 41 MB 100 19 MB None of those statements came near the byte budget, so the whole difference is row count. The floor is a plateau rather than a leak: 1,000,000 rows written at 1000 per statement settles around 229 MB and stops climbing. Adds SPEND_LOG_WRITE_BATCH_MAX_ROWS, default 100, applied alongside the existing byte budget so whichever binds first splits the statement. Both are needed, since bytes are what track a prompt-carrying row and rows are what track the engine's per-row bookkeeping. One consequence worth naming: a flush now issues more statements, and a statement that fails under a poison flood costs one insert before any isolation runs, so the irreducible floor rises by the statement count. The isolation budget still caps the amplification on top of that, and the tests assert the bound derived from the configured row cap rather than a constant.		2026-08-21 09:49:51 -07:00
..
example_config_yaml	test: test	2026-03-28 19:17:38 -07:00
test_configs	test: test	2026-03-28 19:17:38 -07:00
test_model_response_typing	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
azure_fine_tune.jsonl	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
batch_job_results_furniture.jsonl	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
conftest.py	[Fix] conftest: flush cache instances and warn on silent skips	2026-04-20 22:19:36 -07:00
data_map.txt	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
eagle.wav	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
gettysburg.wav	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
large_text.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
messages_with_counts.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
model_cost.json	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
openai_batch_completions.jsonl	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
openai_batch_completions_router.jsonl	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
speech_vertex.mp3	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
test_aproxy_startup.py	(Security fix) - Upgrade to fastapi==0.115.5 (#7447)	2024-12-28 17:08:19 -08:00
test_audit_logs_proxy.py	fix(proxy): require opt in for audit header fallback	2026-04-30 11:17:04 -07:00
test_auth_checks.py	test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769)	2026-08-20 20:24:49 -07:00
test_banned_keyword_list.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
test_blog_posts_endpoint.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_check_batch_cost.py	fix(proxy): claim batch cost rows atomically so multi-pod polling can't double-bill (#37685)	2026-08-20 16:21:04 -07:00
test_check_responses_cost.py	fix(proxy): fall back to the SDK when a queued response's deployment is missing	2026-08-05 23:13:15 -07:00
test_custom_callback_input.py	ci: lint the test tree for undefined names and fix all 30 (#37671)	2026-08-20 13:30:34 -07:00
test_custom_logger_s3_gcs.py	chore(tests): thread config_file_path through s3/gcs custom-logger tests	2026-05-13 01:13:52 +00:00
test_custom_tokenizer_bug.py	test: make custom_tokenizer proxy tests hermetic (#29643)	2026-06-04 12:51:37 -07:00
test_db_schema_changes.py	test: initial test to enforce all functions in user_api_key_auth.py h… (#7797)	2025-01-15 21:52:45 -08:00
test_default_end_user_budget_simple.py	Litellm oss staging 030626 (#29578)	2026-06-03 11:01:51 -07:00
test_deployed_proxy_keygen.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
test_deprecated_key_grace_period.py	Litellm key rotation bug (#27756)	2026-05-12 17:16:37 -07:00
test_e2e_pod_lock_manager.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_gemini_agents_endpoints.py	Gemini managed agents support (#28270)	2026-05-19 16:02:03 -07:00
test_get_favicon.py	test(proxy): align favicon remote asset expectations	2026-04-30 11:46:45 -07:00
test_get_image.py	fix(static-assets): browser-load remote branding assets	2026-04-30 11:30:57 -07:00
test_google_endpoint_routing.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_google_gemini_proxy_request.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_jwt.py	test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769)	2026-08-20 20:24:49 -07:00
test_jwt_key_mapping.py	feat(auth): resolve caller identity once into a Principal at the auth seam (#30887)	2026-06-20 18:49:41 -07:00
test_key_generate_dynamodb.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
test_key_generate_prisma.py	fix(spend-tracking): drop orphaned imports; align tests with alias contract	2026-04-29 18:53:12 +00:00
test_models_fallback_endpoint.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_multipart_bypass_repro.py	fix: harden /key/update authorization checks (#27878)	2026-05-14 04:16:04 +00:00
test_prisma_client_backoff_retry.py	fix(tests): mock prisma.Prisma in backoff retry tests to avoid 'prisma generate'	2026-02-17 19:29:20 -03:00
test_prompt_test_endpoint.py	fix: harden /key/update authorization checks (#27878)	2026-05-14 04:16:04 +00:00
test_proxy_config_unit_test.py	test(lint): ban blind pytest.raises(Exception) with ruff B017 (#37731)	2026-08-20 18:09:42 -07:00
test_proxy_custom_auth.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
test_proxy_custom_logger.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_proxy_encrypt_decrypt.py	test: fix test	2025-07-27 09:52:22 -07:00
test_proxy_exception_mapping.py	[Perf] Embeddings: Use router's O(1) lookup and shared sessions (#16344)	2025-11-14 09:21:45 -08:00
test_proxy_gunicorn.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
test_proxy_pass_user_config.py	test: test	2026-03-28 19:17:38 -07:00
test_proxy_reject_logging.py	[internal copy of #29089] fix: duplicate claude code traces (#29311)	2026-05-29 22:23:24 -07:00
test_proxy_routes.py	chore(ci): merge dev branch (#28801)	2026-05-25 13:44:49 -07:00
test_proxy_server.py	test: enforce PT012 so a pytest.raises block cannot hide dead assertions (#37748)	2026-08-20 19:36:26 -07:00
test_proxy_server_keys.py	fix tests	2025-10-25 10:19:24 -07:00
test_proxy_server_spend.py	fix(pattern_match_deployments.py): default to user input if unable to… (#6632)	2024-11-08 00:55:57 +05:30
test_proxy_setting_guardrails.py	fix(lakera-guardrail): use os.environ.get() to avoid KeyError on missing LAKERA_API_KEY	2026-02-17 19:30:22 -03:00
test_proxy_token_counter.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_proxy_utils.py	test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769)	2026-08-20 20:24:49 -07:00
test_realtime_cache.py	feat: litellm oss 110626 (#30202)	2026-06-11 22:30:26 -07:00
test_reducto_ocr_route.py	Litellm oss staging 04 21 2026 2 (#26569)	2026-05-20 21:25:19 -07:00
test_request_size_limit_middleware.py	Fix early proxy request size enforcement (#27311)	2026-05-06 12:29:11 -07:00
test_response_polling_handler.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_response_polling_pre_call_checks.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_search_api_logging.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_server_root_path.py	fix: server rooth path (#19790)	2026-01-26 09:48:06 -08:00
test_skills_db.py	test(lint): ban blind pytest.raises(Exception) with ruff B017 (#37731)	2026-08-20 18:09:42 -07:00
test_ui_path_detection.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_unit_test_max_model_budget_limiter.py	fix(proxy): make per-model budgets track spend, enforce, and report the same counter (#37736)	2026-08-21 09:47:52 -07:00
test_unit_test_proxy_hooks.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
test_update_daily_tag_spend.py	perf(spend): write each daily spend batch in one upsert statement (#36448)	2026-08-10 17:06:00 -07:00
test_update_spend.py	fix(spend): bound each spend-log write statement by row count as well as bytes (#37758)	2026-08-21 09:49:51 -07:00
test_user_api_key_auth.py	fix(proxy): make per-model budgets track spend, enforce, and report the same counter (#37736)	2026-08-21 09:47:52 -07:00
test_zero_cost_model_budget_bypass.py	style: run black formatter on files from main merge	2026-04-17 13:02:59 -07:00
vertex_key.json	test: update to new vertex ai keys	2026-03-28 20:19:05 -07:00