Re: IP address used by Sockets

[Scott Klement <rpg400-l@xxxxxxxxxxxxxxxx>]

Joe Wells wrote:
> While testing yesterday, the connection timed out.
> Netstat showed that the local ip was bb.bb.bb.bbb.  This is why it would not
> connect.

No.  This isn't why it wouldn't connect.  Instead, it's a symptom of the 
fact that something is wrong with your routing table.  It won't connect 
because the system has chosen the wrong route to get to the remote 
server.  Using the wrong route caused it to pick the wrong network 
interface and therefore IP address.  It also caused it not to connect.


> I tried again this morning and it used aa.aa.aa.aa and it ran fine.
> I will try to get the hospital network folks to open up bb.bb.bb.bbb like
> aa.aa.aa.aa, 

Why would you do that?!  That's your only assurance that it's using the 
correct route (and therefore going through the encrypted tunnel.)  You 
need to FIX your routing so that data goes through the protected tunnel. 
   Your reaction to this is to allow non-protected data?

Unless I'm misunderstanding the situation?

> In theory, I understand what Michael is suggesting.  Unfortunately, I am new
> to sockets and am struggling with how/where to specify the ip address.  The
> bind() that is called in the server program does not specify an ip, only a
> port.

Set sin_addr to the IP address (in binary form) you want to bind to. 
It's not any different from when you set the address in the same 
structure when you use it with connect()