Re: Can't use internal network after dialup modem is used -- get ping: sendto: Operation not permitted

From: Enrique Perez-Terron (enrio_at_online.no)
Date: 10/07/05 

Date: Fri, 07 Oct 2005 17:21:40 +0200

On Fri, 07 Oct 2005 14:52:56 +0200, Jean-David Beyer <jdbeyer@exit109.com> wrote:

> Nico Kadel-Garcia wrote (in part):
>> "Jean-David Beyer" <jdbeyer@exit109.com> wrote in message
>
[...]
>>> # /sbin/route -n
>>> Kernel IP routing table
>>> Destination Gateway Genmask Flags Metric Ref Use Iface
>>> 64.59.194.1 0.0.0.0 255.255.255.255 UH 0 0 0 ppp0
>>> 192.168.2.0 0.0.0.0 255.255.255.0 U 0 0 0 eth1
>>> 192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
>>> 169.254.0.0 0.0.0.0 255.255.0.0 U 0 0 0 eth1
>>> 0.0.0.0 46.99.491.1 0.0.0.0 UG 0 0 0 ppp0
[...]
> The only reason I was concerned about routing is to answer the question,
> "Where is the nameserver told what port to use to resolve hostnames into IP
> addresses?" I.e., why does it not look at my two LAN networks? It does not,

Port = tcp, udp port. You don't mean that, do you?

> and there is no point to trying, but I no longer recall how it knows that.
> So if my machine had two ports that were capable of accessing the Internet,
> how would it decide which to use?

What interface to use? The nameserver uses the socket(), bind(), and connect()
library calls, usually with the argument INADDR_ANY as the local address.
Then the kernel decides, using the routing tables and the connect-to address
how to send out the IP segments.

If <connect-to> is the IP address the name server specifies in the connect()
call, and <dest> and <mask> are the data in the similarly named columns in
the routing table, the kernel steps down the table rows, until it find one
that satisfies

   (<connect-to> & <mask>) == (<dest> & <mask>)

If the line found in this manner has a 'G' flag, the kernel may use
the <gateway> address to determine e.g. what ethernet address to use,
when sending the IP segment to <connect-to> over the <iface>.

If there is not 'G', the <connect-to> is used to determine the
ethernet address, or whatever the next lower network layer requires.

Of course, you could say that this just pushes the question one step
back: Where does the information in the routing table come from?
It comes partly from the configuration files on your computer, and
partly from the ppp setup, where the remote end supplies a couple of
paraters.

-Enrique

Relevant Pages

  • RE: Routing problem
    ... You say you promote the properties in the send port. ... be promoted and BizTalk can use them to route the message. ... the messages gets suspended with a 0c0c01680 Routing Failure: ...
    (microsoft.public.biztalk.general)
  • Re: PC Anywhere
    ... routing and remote access ... >> one for port 5631 point it to the IP address of the work station you want ... I went into Administrative Tools> Routing and Remote Access and, ...
    (microsoft.public.windows.server.sbs)
  • Re: read hang on datagram socket
    ... >> port 0, an ephemeral port is assigned right away. ... > Programming in the UNIX Environment). ... > the routing table for each: ...
    (freebsd-hackers)
  • Re: Etherchannel plus routing at the same time?
    ... Routing is working well. ... There are two identical 100Mbps fiber optic lines and sets of media converters going to the C2950 at the remote site, so what I'd like to know is if it is possible to bond two ports in each switch together in an etherchannel, yet still keep a nailed IP address on the etherchannel at the C3750 switch since it would be acting as like a router's interface port? ... IP address to that vlan, and making the fast ethernet port which is connected to the fiber link into a member of that vlan. ...
    (comp.dcom.sys.cisco)