Reusing Persistent SSH Connections for Performance Reasons

Some weeks ago I had to run a simulation for my research which has been very elaborate and involved externally invoked scripts on remote machines (so-called Wuchtbrummen). These external calls were dispatched by a central main script, that was running on my local PC. To avoid a complete SSH session setup and teardown every time a new call should be invoked on a Wuchtbrumme, I wanted to have a persistent SSH connection, where all subsequent commands can be tunneled through.

I was pretty surpised when I saw this available using standard OpenSSH options. I also posted this on serverfault, even though I was a bit late to get the ‘best answer badge’. I just post the necessary steps here to share it with you and to have it quickly searchable at hand.

The main option to set is the ControlMaster option. To enable it, I add the following line to ~/.ssh/ssh_config

ControlMaster auto

Then, I added the following line to the Host Section in ~/.ssh/ssh_config

Host *
  ControlPath ~/.ssh/master-%r@%h:%p

Now start a master session in the background, which all subsequent SSH sessions will reuse using the following command (substitute remote.host with your remote machine or Wuchtbrumme):

ssh -MNf remotehost

And that’s it again!
Cheers, iss