[CI] DEBUG test develop with Paddle release/3.3 latest build#7588
[CI] DEBUG test develop with Paddle release/3.3 latest build#7588EmmonsCurse wants to merge 3 commits into
Conversation
|
Thanks for your contribution! |
|
/skip-ci ci_iluvatar |
PaddlePaddle-bot
left a comment
There was a problem hiding this comment.
🤖 Paddle-CI-Agent | pr_review |
2026-06-22 18:03:52
📋 Review 摘要
PR 概述:将部分 Linux GPU CI 的 Paddle 安装源改为 Paddle release/3.3 相关 wheel。
变更范围:.github/workflows/_build_linux.yml、.github/workflows/_pre_ce_test.yml
影响面 Tag:[CI]
问题
| 级别 | 文件 | 概述 |
|---|---|---|
| 🟡 建议 | .github/workflows/_build_linux.yml:199 |
安装源固定到单次 wheel,无法持续验证 release/3.3 latest |
| 🟡 建议 | .github/workflows/_pre_ce_test.yml:204 |
仅部分测试 workflow 切到 release/3.3,其他测试仍走 nightly,验证链路不一致 |
📝 PR 规范检查
符合规范。
总体评价
本 PR 的方向与 [CI] 影响面匹配,但当前实现把“latest”固化成了具体历史 wheel,并且没有把同一 Paddle wheel 统一透传到相关测试 workflow。建议先把 Paddle 安装源收敛到可配置的 latest URL,再让构建和测试作业复用同一输入。
| python -m pip install paddlepaddle-gpu==${PADDLEVERSION} -i https://www.paddlepaddle.org.cn/packages/stable/cu126/ | ||
| else | ||
| python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/ | ||
| python -m pip install https://paddle-qa.bj.bcebos.com/paddle-pipeline/Release-TagBuild-Training-Linux-Gpu-Cuda12.6-Cudnn9.5-Trt10.5-Mkl-Avx-Gcc11-SelfBuiltPypiUse/2b9f8b689bc8988f97a5ede056c8c81bfa0332c2/paddlepaddle_gpu-3.3.1.post20260420+2b9f8b689bc-cp310-cp310-linux_x86_64.whl --extra-index-url https://www.paddlepaddle.org.cn/packages/stable/cu126/ |
There was a problem hiding this comment.
🟡 建议 这里写死到了一次具体产物 2b9f8b689bc8988f97a5ede056c8c81bfa0332c2/paddlepaddle_gpu-3.3.1.post20260420...whl,不是 PR 描述里的 release/3.3 latest。
ce_job.yml / publish_job.yml 已经通过 PADDLE_WHL_URL 和 COMPILE_USE_PADDLE_WHL_URL_MAPPINGS 支持分支级 wheel URL;写死后后续 release/3.3 新包不会被覆盖到,CI 会持续验证旧构建。
建议修复方式:把 release/3.3 的 latest wheel URL 放到 PADDLE_WHL_URL/仓库变量中由调用方传入,或改成 release/3.3 的 /latest/ 路径,并保留当前 PADDLE_WHL_URL 优先级。
| git config --global --add safe.directory /workspace/FastDeploy | ||
| cd FastDeploy | ||
| python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/ | ||
| python -m pip install https://paddle-qa.bj.bcebos.com/paddle-pipeline/Release-TagBuild-Training-Linux-Gpu-Cuda12.6-Cudnn9.5-Trt10.5-Mkl-Avx-Gcc11-SelfBuiltPypiUse/2b9f8b689bc8988f97a5ede056c8c81bfa0332c2/paddlepaddle_gpu-3.3.1.post20260420+2b9f8b689bc-cp310-cp310-linux_x86_64.whl --extra-index-url https://www.paddlepaddle.org.cn/packages/stable/cu126/ |
There was a problem hiding this comment.
🟡 建议 这个测试 workflow 也写死同一个 wheel,但调用方只向 _build_linux.yml 传入 PADDLE_WHL_URL,其他测试 workflow 仍安装 nightly/cu126。
这样构建产物与大部分测试作业不是同一 Paddle 来源,不能完整验证 develop + release/3.3;后续切换 wheel 时还需要逐个改 workflow。
建议修复方式:给 _pre_ce_test.yml 及相关测试 workflow 增加 PADDLE_WHL_URL 输入,并由 ci_image_update.yml / ce_job.yml / publish_job.yml 统一透传同一个 release/3.3 latest wheel。
Motivation
Validate compatibility between FastDeploy develop branch and the latest build from Paddle release/3.3. This is used for debugging and verification before making any version alignment or upgrade decisions.
Modifications
Usage or Command
N/A
Accuracy Tests
N/A
Checklist
[FDConfig],[APIServer],[Engine],[Scheduler],[PD Disaggregation],[Executor],[Graph Optimization],[Speculative Decoding],[RL],[Models],[Quantization],[Loader],[OP],[KVCache],[DataProcessor],[BugFix],[Docs],[CI],[Optimization],[Feature],[Benchmark],[Others],[XPU],[HPU],[GCU],[DCU],[Iluvatar],[Metax]]pre-commitbefore commit.releasebranch, make sure the PR has been submitted to thedevelopbranch, then cherry-pick it to thereleasebranch with the[Cherry-Pick]PR tag.