Millet Porridge

English version of https://corvo.myseu.cn

0%

OpenSSH Series 4: Reuse ssh connection

Posted on Edited on In Disqus:
Thanks to the OpenSSH developers, it supports multiplexing which allow us to use an already establihed TCP connection when we want to connect to the remote server. This blog will talk about how to use this feature.

multiplexing or keepalive

When we talk about the HTTP protocol, some users could misunderstand multiplexing and keepalive. While the keepalive means you can handle many requests in one connection, and it’s need FIFO. The multiplexing seems an improvement for keepalive, one connection also holds many requests, but don’t need FIFO.

If you connect the remote server frequently, like git push many times, the multiplexing could help to save your time. Because it doesn’t need to establish a new TCP connection every time.

Configuration

Well, the OpenSSH supports multiplexing, and clients could share the connection with the Unix socket. Just add the Controlpath under the old configuration.

1
2
3
4
5
6
7
8
9
Host ali
    HostName 1.2.3.4
    Port 22
    User root
    IdentityFile ~/.ssh/id_rsa_test

    Controlmaster auto
    Controlpath ~/.ssh/ssh-%r@%h:%p.sock
    ControlPersist 600

When you firstly connect the server, a Unix socket file is automatically created in your ~/.ssh directory.

1
2
3
~ ❤  ls ~/.ssh
...
ssh-root@1.2.3.4:22.sock

This file could be used by other ssh clients who want to connect the same server.

ControlPersist: Please be noted that if the initial client closed, other clients will be closed after that time(600s).

Since we use sshfs, we could combine it with OpenSSH multiplexing. By simply mounting a path, we don’t need to open a new console to stay connected.

Related Posts