Ask HN: moving beyond screen to persist SSH sessions?

    ssh -t kepler bin/irc -D 1083 -L 3128:localhost:3128

    #!/bin/sh

    if [ -f /tmp/jreese.agent.pid -a -f /tmp/jreese.agent.sock ]; then
            echo "Socket and PID files found; loading..."

            export SSH_AGENT_PID=`cat /tmp/jreese.agent.pid`
            export SSH_AUTH_SOCK=`cat /tmp/jreese.agent.sock`
    else
            echo "Socket and PID files not found; starting SSH agent..."

            UMASK=`umask`
            umask 077

            eval `ssh-agent`;
            ssh-add $HOME/.ssh/id_rsa

            echo "Agent PID: $SSH_AGENT_PID"
            echo "Agent Socket: $SSH_AUTH_SOCK"

            echo $SSH_AGENT_PID > /tmp/jreese.agent.pid
            echo $SSH_AUTH_SOCK > /tmp/jreese.agent.sock

            umask $UMASK
    fi

    screen -Rd irssi -c ~/.screenrc.irssi -p 1

This script sets up a persistent, "secure" SSH key agent on the remote machine that can be reused for as many disconnects and reconnects as I feel like making until the server (if ever) needs a reboot. Then it starts or reconnects to a named screen session that I can continue using as if nothing ever happened.

Combine this method with a few different aliases as needed, and a local session key agent to only input your key's passphrase once, and then after a disconnect, press up-enter in each terminal as necessary, and boom, you're back in business.

--

On a related note, a drop of network connectivity should not be killing your SSH connection, especially for as short a time span as one second. SSH is designed to stay connected through at least 30 seconds of network loss; after the connection is regained, it should automatically reconnect to the SSH server and continue the session where it left off, even maintaining a buffer of unsent input during the connection loss.

I've personally suspended and then unsuspended my laptop within a one-or-two minute period without losing my SSH session, althought that's likely stretching things.

What's certain is that a 1-2 second lapse in your connection should most certainly not be dropping your SSH connection; you should definitely be investigating that to figure out what the problem is.

Secondly, you should be able to set scripts to trigger when a network interface comes up in most linux distros, but how to do it varies, for example /etc/network/ifup.d/ in ubuntu.

1. Get an Android.

2. Hack it and install the debian subsystem.

3. see: http://brad.livejournal.com/2400054.html

So now instead of running an ssh to a remote server you are running an ssh to your android which then hosts all of the outward ssh sessions. Because of the androids 2g/3g you can't lose your connection (well not easily).

From the ssh_config manpage:

"TCPKeepAlive

Specifies whether the system should send TCP keepalive messages to the other side. If they are sent, death of the connection or crash of one of the machines will be properly noticed. However, this means that connections will die if the route is down temporarily, and some people find it annoying.

The default is ``yes'' (to send TCP keepalive messages), and the client will notice if the network goes down or the remote host dies. This is important in scripts, and many users want it too.

To disable TCP keepalive messages, the value should be set to ``no''."