Re: ssh ControlMaster problem, workaround

[Stefan Sperling]
> And the best long term fix for that is to convince the OpenSSH
> developers to include this patch. Or something comparable, such
> as a command line flag that prevents output from signal handlers.
>
> Have you talked to upstream about this yet?

I have not - I don't maintain OpenSSH for Debian. Once the maintainer
applied that patch (on my recommendation), I patched subversion as I
indicated, and didn't think much more about it.

I'm CCing Colin Watson, who applied the patch (the fatal() patch in
openssh 1:4.2p1-6). Colin, have you forwarded the patch upstream?

> Another idea:
> Couldn't we redirect stderr of the ssh process to somewhere else
> then the screen before we kill it?

Well, ssh could print many things to stderr, including other connection
errors, or password prompts, or errors forwarded from the other end of
the tunnel. I don't think we want to suppress those. (And you can't
redirect the child's stderr after you exec("ssh").)

-- 
Peter Samuelson | org-tld!p12n!peter | http://p12n.org/