From e7f3afe2e7176625935b7a153960aa02fefc84f2 Mon Sep 17 00:00:00 2001 From: Yuneng Jiang Date: Fri, 24 Apr 2026 22:59:24 -0700 Subject: [PATCH] [Infra] add worker restart and retry flags to local_testing CI jobs Match the resilience config that the llm_translation job already uses: --max-worker-restart=5 lets xdist replace workers killed by pytest-timeout's os._exit(1) instead of failing the whole job, and --retries 2 --retry-delay 5 self-heals transient network flakes. Preserves real integration test coverage; no test code changes needed. --- .circleci/config.yml | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/.circleci/config.yml b/.circleci/config.yml index 9f01bae3e5b..62aca425582 100644 --- a/.circleci/config.yml +++ b/.circleci/config.yml @@ -235,7 +235,10 @@ jobs: -k \"not test_python_38.py and not test_basic_python_version.py and not router and not assistants and not langfuse and not caching and not cache\" \ -n 4 \ --timeout=300 \ - --timeout_method=thread" + --timeout_method=thread \ + --retries 2 \ + --retry-delay 5 \ + --max-worker-restart=5" no_output_timeout: 15m - run: name: Rename the coverage files @@ -300,7 +303,10 @@ jobs: -k \"not test_python_38.py and not test_basic_python_version.py and not router and not assistants and not langfuse and not caching and not cache\" \ -n 4 \ --timeout=300 \ - --timeout_method=thread" + --timeout_method=thread \ + --retries 2 \ + --retry-delay 5 \ + --max-worker-restart=5" no_output_timeout: 15m - run: name: Rename the coverage files