From 2e2140fdb12ea859eda32cc20fb7f707701ca398 Mon Sep 17 00:00:00 2001
From: youkaichao <youkaichao@gmail.com>
Date: Fri, 21 Jun 2024 14:04:26 -0700
Subject: [PATCH] [ci][test] fix ca test in main (#5746)

---
 .buildkite/test-pipeline.yaml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/.buildkite/test-pipeline.yaml b/.buildkite/test-pipeline.yaml
index c337a81d4a0d2..0b87e6280f0bb 100644
--- a/.buildkite/test-pipeline.yaml
+++ b/.buildkite/test-pipeline.yaml
@@ -197,6 +197,9 @@ steps:
   gpu: a100
   num_gpus: 4
   commands: 
+  # FIXIT: find out which code initialize cuda before running the test
+  # before the fix, we need to use spawn to test it
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
   # NOTE: don't test llama model here, it seems hf implementation is buggy
   # see https://github.com/vllm-project/vllm/pull/5689 for details
   - pytest -v -s distributed/test_custom_all_reduce.py