Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

分布式集群方式部署,只要有1个worker挂了,整个集群都无法正常工作 #2585

Open
1 of 3 tasks
swy0915 opened this issue Nov 26, 2024 · 0 comments
Open
1 of 3 tasks
Labels
Milestone

Comments

@swy0915
Copy link

swy0915 commented Nov 26, 2024

System Info / 系統信息

centos7

Running Xinference with Docker? / 是否使用 Docker 运行 Xinfernece?

  • docker / docker
  • pip install / 通过 pip install 安装
  • installation from source / 从源码安装

Version info / 版本信息

docker镜像 xinference v0.16.3

The command used to start Xinference / 用以启动 xinference 的命令

集群部署:
master:
docker run -itd --shm-size=4g --ulimit memlock=-1 -v /data2/sang/models:/models -v /data2/sang/xinference:/xinference -e XINFERENCE_HOME=/xinference -p 6000:6000 --name xinference-master --gpus all --net=host registry.cn-hangzhou.aliyuncs.com/xprobe_xinference/xinference:v0.16.3 xinference-supervisor -H 172.20.107.176 --port 6000

worker1:
docker run -itd --shm-size=4g --ulimit memlock=-1 -v /data2/sang/models:/models -v /data2/sang/xinference:/xinference -e XINFERENCE_HOME=/xinference -p 6001:6001 --name xinference-worker --gpus all --net=host registry.cn-hangzhou.aliyuncs.com/xprobe_xinference/xinference:v0.16.3 xinference-worker -e "http://172.20.107.176:6000" -H 172.20.107.176 --worker-port 6001

worker2:
docker run -itd --shm-size=4g --ulimit memlock=-1 -v /data2/sang/models:/models -v /data2/sang/xinference:/xinference -e XINFERENCE_HOME=/xinference -p 6001:6001 --name xinference-worker --gpus all --net=host registry.cn-hangzhou.aliyuncs.com/xprobe_xinference/xinference:v0.16.3 xinference-worker -e "http://172.20.107.176:6000" -H 172.20.107.175 --worker-port 6001

Reproduction / 复现过程

运行任何模型,replica为2 (这样每个worker都会存一份),然后停止其中一个worker(docker stop xinference-worker),但是另1个worker是好的,但是整个集群都无法正常工作,模型无法提供服务。

Expected behavior / 期待表现

我发现已经有其他用户提过此问题,希望能够继续提供服务,因为另1个worker还存活着,而且也是加载了这个模型的。

@XprobeBot XprobeBot added the gpu label Nov 26, 2024
@XprobeBot XprobeBot added this to the v1.x milestone Nov 26, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

2 participants