Port forwarding oddity - SSH

This is a discussion on Port forwarding oddity - SSH ; I have a *really* odd port forwarding problem. Connections work properly between any two machines at the same physical location, but hang between any two machines at different locations. All hardware and software is identical - ssh and sshd are ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Port forwarding oddity

  1. Port forwarding oddity

    I have a *really* odd port forwarding problem. Connections work properly
    between any two machines at the same physical location, but hang between
    any two machines at different locations. All hardware and software is
    identical - ssh and sshd are 3.7.1p2 running on Sparc Solaris 8.

    Regular ssh connections work fine, both for interactive logins and
    remote command execution. But using port forwarding via

    ../ssh -v -v -v -f -g -L 2202:machine1.remote.invalid:25 \
    machine1.local.invalid tail -f /dev/null

    the forwarding never gets established. It hangs at this point of the
    debug output:

    debug1: Connection to port 2202 forwarding to machine1.remote.invalid
    port 25 requested.
    debug1: fd 6 setting TCP_NODELAY
    debug2: fd 6 is O_NONBLOCK
    debug2: fd 6 is O_NONBLOCK
    debug1: channel 3: new [direct-tcpip]

    and then it simply stops until I control-] out of it. However, between
    machines at any of the same locations, it continues on to

    debug1: fd 6 setting TCP_NODELAY
    debug2: fd 6 is O_NONBLOCK
    debug2: fd 6 is O_NONBLOCK
    debug1: channel 3: new [direct-tcpip]
    debug1: channel 3: open confirm rwindow 131072 rmax 32768

    and I get the SMTP prompt. This smells like a network problem, but I
    can't think of anything that would affect port forwarding without
    affecting other ssh connections. Any ideas would be *greatly* welcome.

    Thanks,
    JS

  2. Re: Port forwarding oddity


  3. Re: Port forwarding oddity

    Richard E. Silverman wrote:
    > Perhaps http://www.snailbook.com/faq/mtu-mismatch.auto.html


    Unfortunately, it didn't help. Symptoms and debug output remained
    identical...

    JS

+ Reply to Thread