PaddlePaddle / PaddlePaddle/docs
paddlepaddle分布式docker部署命令行错误
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 284
- Forks
- 906
- PR merge metrics
- No merged PRs in 30d
Description
文档链接&描述 Document Links & Description
文档链接:https://www.paddlepaddle.org.cn/documentation/docs/zh/guides/06_distributed_training/deployment_cn.html
错误位置:裸机及Docker化部署-paddle环境安装-docker安装命令
错误原因:docker --host命令部分版本不支持,本机20.10.12版本测试不通过,报错
unknown flag: --host
同时,由于没有未挂载/dev/shm会导致多卡无法正常运行,只能正常运行一卡。
更多建议:nvidia-docker2貌似已经弃用nvidia-docker命令,改为--runtime方式,可以在命令行中进行体现,同时,在启动docker时将/dev/shm进行挂载。
请提出你的建议/提个PR修复这个问题 Please give your suggestion/update a PR to fix it
命令行代码可以修改为:
docker run --runtime=nvidia --gpus all --name paddle -it --network host -v $PWD:/paddle -v /dev/shm:/dev/shm registry.baidubce.com/paddlepaddle/paddle:2.2.2-gpu-cuda11.2-cudnn8 /bin/bash
--network host:使用主机网络-v /dev/shm:/dev/shm:挂载共享内存目录以便多卡可以正常启动
本机实际测试通过,附参考连接
另附nvidia container toolkit安装教程
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Open the linked distributed-training deployment document and find the bare-metal and Docker deployment section under paddle environment installation. Update the Docker command to use the proposed runtime and GPU options, host networking, and /dev/shm mount; done means the documented command no longer uses the unsupported --host flag and supports multi-GPU startup.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- docker
- Domain
- documentation
- Issue type
- Documentation
- Difficulty
- 1/5
- Estimated time
- Under an hour
- Activity status
- Stale
- Clarity
- Clearly specified
- Newbie friendliness
- 55/100