Last year I wrote some Korn shell scripts to control various programs on our hosts. It consists of a menu script on the "parent" host that makes SSH calls to "child" scripts on the remote hosts. This allows us to stop, start & status the remote programs without having to manually log in to each of the remote hosts. The SSH calls are similar to this:
ssh remoteuser@remotehost STATUS arg2 arg3 ...
where "STATUS arg2 arg3 ..." are arguments to the child script on remotehost.
For security reasons, the child program is not specified in the SSH call. Instead its given in the entry in the authorized_keys file in the .ssh directory on remotehost. It looks like this:
no-port-forwarding,no-X11-forwarding,no-agent-forwarding,command="${HOME}/bin/remote_control.ksh ${SSH_ORIGINAL_COMMAND}" ssh-rsa verylongkey= user@parenthost
where ${HOME}/bin/remote_control.ksh is the name of the child program on remotehost
and ${SSH_ORIGINAL_COMMAND} is given the arguments passed in the SSH call.
After the previous example call, ${SSH_ORIGINAL_COMMAND} would be equal to "STATUS arg2 arg3 ...".
The Korn shell scripts work but have lots of limitations. Recently, I discovered wxPerl and I'm in the process of rewriting the parent program as a GUI. I've also found the Net::SSH2 package and would like to make use of it. The docs show how to log in using user id and passwords, but not how to connect as we do in our menu script. Does anyone have an example?
