Daniel Stenberg wrote:
> On Mon, 7 Sep 2009, Dmitry V. Krivenok wrote:
>
>> I think that it's very bad.
>
> I disagree. You've asked about what libssh2 (in its current form)
> supports and it does support most of what you ask already now.
>
> Most of your problems are however not related to libssh2 or even SSH2
> but are more generic concerns on how you start processes in unix-like
> operating systems, and those are matters we cannot fix in this project.
I think that the only thing libssh2 should support is signals delivery
(section 6.9 of rfc4254).
Having this facility users may implement their own management of remote
processes.
>
>> Maybe sshd is not properly configured, maybe it's not responsible for
>> termination of spawn process (what does RFC say about this?).
>
> It starts a shell in the remote end, I don't think it has much to do
> with what the SECSH RFC says. Read up and tell me if I'm wrong!
I didn't find anything related to this in RFC.
My experiments shown:
1) If I start remote shell (shell channel), execute long running command
and kill client program, then sshd sends SIGHUP (default action is
terminate) to the process.
2) If I start remote command explicitly (exec channel) and kill client
program, then sshd doesn't send any signals to the process.
>
_______________________________________________
libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
Received on 2009-09-07