minor fix

Signed-off-by: Yongbozzz <yongbo.zhu@intel.com>

Author: Yongbozzz

EdgeCraftRAG/docker_compose/intel/gpu/arc/set_env.sh

@@ -22,10 +22,10 @@ export VLLM_SERVICE_PORT_B60=${VLLM_SERVICE_PORT_B60}
 export VLLM_SERVICE_PORT_A770=${VLLM_SERVICE_PORT_A770}
 export TENSOR_PARALLEL_SIZE=${TENSOR_PARALLEL_SIZE}
 
-export SELECTED_XPU_0=${SELECTED_XPU_0}
 export vLLM_ENDPOINT=${vLLM_ENDPOINT}
 export MAX_NUM_SEQS=${MAX_NUM_SEQS}
 export MAX_NUM_BATCHED_TOKENS=${MAX_NUM_BATCHED_TOKENS}
 export MAX_MODEL_LEN=${MAX_MODEL_LEN}
 export LOAD_IN_LOW_BIT=${LOAD_IN_LOW_BIT}
 export CCL_DG2_USM=${CCL_DG2_USM}
+export ZE_AFFINITY_MASK=${ZE_AFFINITY_MASK}

EdgeCraftRAG/tests/test_compose_vllm_on_arc.sh

@@ -29,7 +29,6 @@ UI_UPLOAD_PATH="$WORKPATH/tests"
 HF_ENDPOINT=https://hf-mirror.com
 VLLM_SERVICE_PORT_A770=8086
 TENSOR_PARALLEL_SIZE=1
-SELECTED_XPU_0=0
 vLLM_ENDPOINT="http://${HOST_IP}:${VLLM_SERVICE_PORT_A770}"
 LLM_MODEL="Qwen/Qwen3-8B"
 LLM_MODEL_PATH="${MODEL_PATH}/${LLM_MODEL}"

EdgeCraftRAG/tests/test_compose_vllm_on_arc_b60.sh

@@ -33,6 +33,7 @@ vLLM_ENDPOINT="http://${HOST_IP}:${VLLM_SERVICE_PORT_B60}"
 LLM_MODEL="Qwen/Qwen3-8B"
 VLLM_IMAGE_TAG="1.1-preview"
 DP=1
+ZE_AFFINITY_MASK=1
 
 function build_docker_images() {
     opea_branch=${opea_branch:-"main"}
@@ -62,7 +63,7 @@ function start_services() {
     n=0
     until [[ "$n" -ge 100 ]]; do
         docker logs ipex-serving-xpu-container > ${LOG_PATH}/ipex-serving-xpu-container.log 2>&1
-        if grep -q "Starting vLLM API server on http://0.0.0.0:" ${LOG_PATH}/ipex-serving-xpu-container.log; then
+        if grep -q "Starting vLLM API server" ${LOG_PATH}/ipex-serving-xpu-container.log; then
             break
         fi
         sleep 6s