docker-compose规模独立的卷 [英] docker-compose scale with separate volumes

问题描述

我想使用scale，但是每个节点需要不同的卷和端口映射。

I would like to use scale but I need different volumes and port mapping for each node.

我该怎么做？理想情况下，我需要某种环境变量或运行脚本来为每个新实例分配卷和端口。

How can I do that? Ideally I would need some kind of environment variable or a script that is run to allocate a volume and ports for each new instance.

推荐的方法是什么？

推荐答案

我知道这可能不是一个完美的解决方案，但是如果您不介意在节点之间共享数据的话。做得很好。我正在使用它进行本地测试，因此对于我来说是安全的。

I am aware that this might not be a perfect solution but if you don't mind sharing data between nodes.. it does work well. I am using this for local tests so it is safe in my case.

Docker-compose.yml

Docker-compose.yml

...
volumes:
   - /var/run/docker.sock:/var/run/docker.sock
...

Dockerfile

Dockerfile

...
RUN pip3 install docker
...

节点，我从docker import部署以下脚本 get_name.py

In each node, I deploy the following script get_name.py

from docker import Client
import os

hostname = os.environ['HOSTNAME']

cli = Client(base_url='unix://var/run/docker.sock')
data = cli.containers()

for d in data:
    id = d['Id'][:12]
    names = d['Names']
    if id == hostname:
        print(names[0])
        quit()

print(hostname)

当节点启动时（ start.sh ），它会查询它的名称并创建到相应子目录的符号链接：

And when the node starts (start.sh), it queries its name and creates a symlink to the corresponding subdirectory:

...
NODE_NAME=$(python /root/scripts/get_name.py)
OWN_VOLUME_NAME="/shared_volumes${NODE_NAME}"
ln -s ${OWN_VOLUME_NAME} /data
...

这篇关于docker-compose规模独立的卷的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！