diff --git a/.github/workflows/vllm_ascend_doctest.yaml b/.github/workflows/vllm_ascend_doctest.yaml new file mode 100644 index 0000000..63ba746 --- /dev/null +++ b/.github/workflows/vllm_ascend_doctest.yaml @@ -0,0 +1,102 @@ +# +# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# This file is a part of the vllm-ascend project. +# + +name: 'e2e test / doctest' + +on: + workflow_dispatch: + pull_request: + branches: + - 'main' + - '*-dev' + paths: + # If we are changing the doctest we should do a PR test + - '.github/workflows/vllm_ascend_doctest.yaml' + - 'tests/e2e/doctests/**' + - 'tests/e2e/common.sh' + - 'tests/e2e/run_doctests.sh' + schedule: + # Runs every 4 hours + - cron: '0 */4 * * *' + +# Bash shells do not use ~/.profile or ~/.bashrc so these shells need to be explicitly +# declared as "shell: bash -el {0}" on steps that need to be properly activated. +# It's used to activate ascend-toolkit environment variables. +defaults: + run: + shell: bash -el {0} + +jobs: + test: + strategy: + # Each version should be tested + fail-fast: false + matrix: + vllm_verison: [main, v0.7.3-dev, main-openeuler, v0.7.3-dev-openeuler] + name: vLLM Ascend test + runs-on: linux-arm64-npu-1 + container: + image: m.daocloud.io/quay.io/ascend/vllm-ascend:${{ matrix.vllm_verison }} + steps: + - name: Check NPU/CANN and git info + run: | + echo "====> Print NPU/CANN info" + npu-smi info + cat /usr/local/Ascend/ascend-toolkit/latest/"$(uname -i)"-linux/ascend_toolkit_install.info + + echo "====> Print vllm-ascend git info" + cd /vllm-workspace/vllm-ascend + git --no-pager log -1 || true + echo "====> Print vllm git info" + cd /vllm-workspace/vllm + git --no-pager log -1 || true + + - name: Config OS mirrors - Ubuntu + if: ${{ !endsWith(matrix.vllm_verison, '-openeuler') }} + run: | + sed -i 's|ports.ubuntu.com|mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list + apt-get update -y + apt install git curl -y + + - name: Config OS mirrors - openEuler + if: ${{ endsWith(matrix.vllm_verison, '-openeuler') }} + run: | + yum update -y + yum install git curl -y + + - name: Config pip mirrors + run: | + pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple + + - name: Checkout vllm-project/vllm-ascend repo + uses: actions/checkout@v4 + + - name: Run vllm-ascend/tests/e2e/run_doctests.sh + run: | + # PWD: /__w/vllm-ascend/vllm-ascend + # Address old branch like v0.7.3: + if [ ! -d /vllm-workspace/vllm-ascend/tests/e2e ]; then + echo "Warning: the doctest path doesn't exists, copy now" + cp -r tests/e2e /vllm-workspace/vllm-ascend/tests/ + fi + + # Simulate container to enter directory + cd /workspace + + # Run real test + echo "Test:" + /vllm-workspace/vllm-ascend/tests/e2e/run_doctests.sh diff --git a/docs/source/quick_start.md b/docs/source/quick_start.md index 7ed2ae9..b9641c2 100644 --- a/docs/source/quick_start.md +++ b/docs/source/quick_start.md @@ -68,6 +68,7 @@ The default workdir is `/workspace`, vLLM and vLLM Ascend code are placed in `/v You can use Modelscope mirror to speed up download: + ```bash export VLLM_USE_MODELSCOPE=true ``` @@ -81,6 +82,7 @@ With vLLM installed, you can start generating texts for list of input prompts (i Try to run below Python script directly or use `python3` shell to generate texts: + ```python from vllm import LLM, SamplingParams @@ -108,6 +110,7 @@ vLLM can also be deployed as a server that implements the OpenAI API protocol. R the following command to start the vLLM server with the [Qwen/Qwen2.5-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct) model: + ```bash # Deploy vLLM server (The first run will take about 3-5 mins (10 MB/s) to download models) vllm serve Qwen/Qwen2.5-0.5B-Instruct & @@ -125,12 +128,14 @@ Congratulations, you have successfully started the vLLM server! You can query the list the models: + ```bash curl http://localhost:8000/v1/models | python3 -m json.tool ``` You can also query the model with input prompts: + ```bash curl http://localhost:8000/v1/completions \ -H "Content-Type: application/json" \ @@ -145,10 +150,10 @@ curl http://localhost:8000/v1/completions \ vLLM is serving as background process, you can use `kill -2 $VLLM_PID` to stop the background process gracefully, it's equal to `Ctrl-C` to stop foreground vLLM process: + ```bash -ps -ef | grep "/.venv/bin/vllm serve" | grep -v grep -VLLM_PID=`ps -ef | grep "/.venv/bin/vllm serve" | grep -v grep | awk '{print $2}'` -kill -2 $VLLM_PID + VLLM_PID=$(pgrep -f "vllm serve") + kill -2 "$VLLM_PID" ``` You will see output as below: diff --git a/tests/e2e/common.sh b/tests/e2e/common.sh new file mode 100644 index 0000000..dd20d70 --- /dev/null +++ b/tests/e2e/common.sh @@ -0,0 +1,51 @@ +# bash fonts colors +cyan='\e[96m' +yellow='\e[33m' +red='\e[31m' +none='\e[0m' + +_cyan() { echo -e "${cyan}$*${none}"; } +_yellow() { echo -e "${yellow}$*${none}"; } +_red() { echo -e "${red}$*${none}"; } + +_info() { _cyan "Info: $*"; } +_warn() { _yellow "Warn: $*"; } +_err() { _red "Error: $*" && exit 1; } + +CURL_TIMEOUT=1 +CURL_COOLDOWN=5 +CURL_MAX_TRIES=120 + +function wait_url_ready() { + local serve_name="$1" + local url="$2" + i=0 + while true; do + _info "===> Waiting for ${serve_name} to be ready...${i}s" + i=$((i + CURL_COOLDOWN)) + set +e + curl --silent --max-time "$CURL_TIMEOUT" "${url}" >/dev/null + result=$? + set -e + if [ "$result" -eq 0 ]; then + break + fi + if [ "$i" -gt "$CURL_MAX_TRIES" ]; then + _info "===> \$CURL_MAX_TRIES exceeded waiting for ${serve_name} to be ready" + return 1 + fi + sleep "$CURL_COOLDOWN" + done + _info "===> ${serve_name} is ready." +} + +function wait_for_exit() { + local VLLM_PID="$1" + while kill -0 "$VLLM_PID"; do + _info "===> Wait for ${VLLM_PID} to exit." + sleep 1 + done + _info "===> Wait for ${VLLM_PID} to exit." +} + +SCRIPT_DIR=$(cd "$(dirname "$0")" && pwd) diff --git a/tests/e2e/doctests/001-quickstart-test.sh b/tests/e2e/doctests/001-quickstart-test.sh new file mode 100755 index 0000000..0e59255 --- /dev/null +++ b/tests/e2e/doctests/001-quickstart-test.sh @@ -0,0 +1,55 @@ +#!/bin/bash + +# +# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# This file is a part of the vllm-ascend project. +# + +function simple_test() { + # Do real import test + python3 -c "import vllm; print(vllm.__version__)" +} + +function quickstart_offline_test() { + export VLLM_USE_MODELSCOPE=true + # Do real script test + python3 "${SCRIPT_DIR}/../../examples/offline_inference_npu.py" +} + +function quickstart_online_test() { + export VLLM_USE_MODELSCOPE=true + vllm serve Qwen/Qwen2.5-0.5B-Instruct & + wait_url_ready "vllm serve" "localhost:8000/v1/models" + # Do real curl test + curl http://localhost:8000/v1/completions \ + -H "Content-Type: application/json" \ + -d '{ + "model": "Qwen/Qwen2.5-0.5B-Instruct", + "prompt": "Beijing is a", + "max_tokens": 5, + "temperature": 0 + }' | python3 -m json.tool + VLLM_PID=$(pgrep -f "vllm serve") + _info "===> Try kill -2 ${VLLM_PID} to exit." + kill -2 "$VLLM_PID" + wait_for_exit "$VLLM_PID" +} + +_info "====> Start simple_test" +simple_test +_info "====> Start quickstart_offline_test" +quickstart_offline_test +_info "====> Start quickstart_online_test" +quickstart_online_test diff --git a/tests/e2e/run_doctests.sh b/tests/e2e/run_doctests.sh new file mode 100755 index 0000000..ee6613a --- /dev/null +++ b/tests/e2e/run_doctests.sh @@ -0,0 +1,27 @@ +#!/bin/bash + +# +# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# This file is a part of the vllm-ascend project. +# + +set -eo errexit + +. $(dirname "$0")/common.sh + +_info "====> Start Quickstart test" +. "${SCRIPT_DIR}/doctests/001-quickstart-test.sh" + +_info "Doctest passed."