openssh unix dev - Jul 2001 - connection attempt timeout - possible feature?

This was talked about on efnet #unixhelp about a week ago while talking
about why ssh takes such a long time to timeout.  It's a feature of the OS
and the connect() function call in question, IIRC.

So in theory one could wrap the call in sshconnect.c (both places) with a
timer to interrupt connect().  Not sure if there is a cleaner way off
hand.

- Ben

On Wed, 4 Jul 2001 carl at bl.echidna.id.au wrote:
>
> I'm writing a network monitoring script, that has to traverse
> a bunch of firewalls (joy ...) and so for connectivity tests,
> I'm using ssh (OpenSSH 2.x).
>
> I'm being lazy, and trying to avoid parallel tasks/threading,
> so I'm interested in reducing the amount of time OpenSSH will wait
> to connect to something before giving up.  I've already set the
> ConnectionAttempts flag to 1, but am interested in reducing the amount of
> time it uses initially. From our boxes, the timeout seems to be abuot 60
seconds,
> but I haven't timed it yet, or groveled through the code enough
> to know for sure.
>
>
> Before I embark on a coding spree and hack at OpenSSH to
> do this, am I looking in the right place?  Is this behaviour
> controlled by the application or the OS?  My copy of Stevens is
> at home, and I'm not, so my usual reference isn't available right
now.
>
> If it is an application specific, would the OpenSSH team be interested in
> a patch if I write one?
>
> Carl
>
>

I'm writing a network monitoring script, that has to traverse 
a bunch of firewalls (joy ...) and so for connectivity tests,
I'm using ssh (OpenSSH 2.x).

I'm being lazy, and trying to avoid parallel tasks/threading,
so I'm interested in reducing the amount of time OpenSSH will wait
to connect to something before giving up.  I've already set the
ConnectionAttempts flag to 1, but am interested in reducing the amount of
time it uses initially. From our boxes, the timeout seems to be abuot 60
seconds,
but I haven't timed it yet, or groveled through the code enough
to know for sure.  


Before I embark on a coding spree and hack at OpenSSH to 
do this, am I looking in the right place?  Is this behaviour 
controlled by the application or the OS?  My copy of Stevens is 
at home, and I'm not, so my usual reference isn't available right now.

If it is an application specific, would the OpenSSH team be interested in
a patch if I write one?

Carl