From a83a0f92b56b71855dc38e8e3d9809619e58bcd1 Mon Sep 17 00:00:00 2001
From: yarongmu-google <150371854+yarongmu-google@users.noreply.github.com>
Date: Thu, 8 May 2025 10:20:54 -0700
Subject: [PATCH] [Test] Attempt all TPU V1 tests, even if some of them fail.
 (#17334)

Signed-off-by: Yarong Mu <ymu@google.com>
---
 .../scripts/hardware_ci/run-tpu-v1-test.sh    | 104 +++++++++++++-----
 1 file changed, 74 insertions(+), 30 deletions(-)

diff --git a/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh b/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
index cae1bffe6a3a..939daddad92b 100755
--- a/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
@@ -1,6 +1,6 @@
 #!/bin/bash
 
-set -xue
+set -xu
 
 # Build the docker image.
 docker build -f docker/Dockerfile.tpu -t vllm-tpu .
@@ -24,36 +24,80 @@ docker run --privileged --net host --shm-size=16G -it \
     && export VLLM_XLA_CHECK_RECOMPILATION=1 \
     && echo HARDWARE \
     && tpu-info \
-    && echo TEST_0 \
-    && pytest -v -s /workspace/vllm/tests/v1/tpu/test_perf.py \
-    && echo TEST_1 \
-    && pytest -v -s /workspace/vllm/tests/tpu/test_compilation.py \
-    && echo TEST_2 \
-    && pytest -v -s /workspace/vllm/tests/v1/tpu/test_basic.py \
-    && echo TEST_3 \
-    && pytest -v -s /workspace/vllm/tests/entrypoints/llm/test_accuracy.py::test_lm_eval_accuracy_v1_engine \
-    && echo TEST_4 \
-    && pytest -s -v /workspace/vllm/tests/tpu/test_quantization_accuracy.py \
-    && echo TEST_5 \
-    && python3 /workspace/vllm/examples/offline_inference/tpu.py \
-    && echo TEST_6 \
-    && pytest -s -v /workspace/vllm/tests/v1/tpu/worker/test_tpu_model_runner.py \
-    && echo TEST_7 \
-    && pytest -s -v /workspace/vllm/tests/v1/tpu/test_sampler.py \
-    && echo TEST_8 \
-    && pytest -s -v /workspace/vllm/tests/v1/tpu/test_topk_topp_sampler.py \
-    && echo TEST_9 \
-    && pytest -s -v /workspace/vllm/tests/v1/tpu/test_multimodal.py \
-    && echo TEST_10 \
-    && pytest -s -v /workspace/vllm/tests/v1/tpu/test_pallas.py \
-    && echo TEST_11 \
-    && pytest -s -v /workspace/vllm/tests/v1/entrypoints/llm/test_struct_output_generate.py \
-    && echo TEST_12 \
-    && pytest -s -v /workspace/vllm/tests/tpu/test_moe_pallas.py" \
+    && { \
+        echo TEST_0: Running test_perf.py; \
+        pytest -s -v /workspace/vllm/tests/tpu/test_perf.py; \
+        echo TEST_0_EXIT_CODE: \$?; \
+    } & \
+    && { \
+        echo TEST_1: Running test_compilation.py; \
+        pytest -s -v /workspace/vllm/tests/tpu/test_compilation.py; \
+        echo TEST_1_EXIT_CODE: \$?; \
+    } & \
+    { \
+        echo TEST_2: Running test_basic.py; \
+        pytest -s -v /workspace/vllm/tests/v1/tpu/test_basic.py; \
+        echo TEST_2_EXIT_CODE: \$?; \
+    } & \
+    { \
+        echo TEST_3: Running test_accuracy.py::test_lm_eval_accuracy_v1_engine; \
+        pytest -s -v /workspace/vllm/tests/entrypoints/llm/test_accuracy.py::test_lm_eval_accuracy_v1_engine; \
+        echo TEST_3_EXIT_CODE: \$?; \
+    } & \
+    { \
+        echo TEST_4: Running test_quantization_accuracy.py; \
+        pytest -s -v /workspace/vllm/tests/tpu/test_quantization_accuracy.py; \
+        echo TEST_4_EXIT_CODE: \$?; \
+    } & \
+    { \
+        echo TEST_5: Running examples/offline_inference/tpu.py; \
+        python3 /workspace/vllm/examples/offline_inference/tpu.py; \
+        echo TEST_5_EXIT_CODE: \$?; \
+    } & \
+    { \
+        echo TEST_6: Running test_tpu_model_runner.py; \
+        pytest -s -v /workspace/vllm/tests/tpu/worker/test_tpu_model_runner.py; \
+        echo TEST_6_EXIT_CODE: \$?; \
+    } & \
+    && { \
+        echo TEST_7: Running test_sampler.py; \
+        pytest -s -v /workspace/vllm/tests/v1/tpu/test_sampler.py; \
+        echo TEST_7_EXIT_CODE: \$?; \
+    } & \
+    && { \
+        echo TEST_8: Running test_topk_topp_sampler.py; \
+        pytest -s -v /workspace/vllm/tests/v1/tpu/test_topk_topp_sampler.py; \
+        echo TEST_8_EXIT_CODE: \$?; \
+    } & \
+    && { \
+        echo TEST_9: Running test_multimodal.py; \
+        pytest -s -v /workspace/vllm/tests/v1/tpu/test_multimodal.py; \
+        echo TEST_9_EXIT_CODE: \$?; \
+    } & \
+    && { \
+        echo TEST_10: Running test_pallas.py; \
+        pytest -s -v /workspace/vllm/tests/v1/tpu/test_pallas.py; \
+        echo TEST_10_EXIT_CODE: \$?; \
+    } & \
+    && { \
+        echo TEST_11: Running test_struct_output_generate.py; \
+        pytest -s -v /workspace/vllm/tests/v1/entrypoints/llm/test_struct_output_generate.py; \
+        echo TEST_11_EXIT_CODE: \$?; \
+    } & \
+    && { \
+        echo TEST_12: Running test_moe_pallas.py; \
+        pytest -s -v /workspace/vllm/tests/tpu/test_moe_pallas.py; \
+        echo TEST_12_EXIT_CODE: \$?; \
+    } & \
     # Disable the TPU LoRA tests until the feature is activated
-    # && echo TEST_13 \
-    # && pytest -s -v /workspace/vllm/tests/tpu/lora/" \
-
+    # && { \
+    #     echo TEST_13: Running test_moe_pallas.py; \
+    #     pytest -s -v /workspace/vllm/tests/tpu/lora/; \
+    #     echo TEST_13_EXIT_CODE: \$?; \
+    # } & \
+    wait \
+    && echo 'All tests have attempted to run. Check logs for individual test statuses and exit codes.' \
+"
 
 # TODO: This test fails because it uses RANDOM_SEED sampling
 # && VLLM_USE_V1=1 pytest -v -s /workspace/vllm/tests/tpu/test_custom_dispatcher.py \