My apologies to everyone. This is a pf problem -- I've sorted it out.

Thanks,

-Stephen-



Stephen Bosch wrote:
> jared r r spiegel wrote:
>> On Mon, Jul 17, 2006 at 05:25:38PM -0600, Stephen Bosch wrote:
>>> route add -host 192.168.0.57 -interface enc0
>>>
>>> I get this response:
>>>
>>> route: enc0: bad address
>>   -interface actually takes an address:
>>
>> ---
>>      If the destination is directly reachable via an interface requiring no
>>      intermediary system to act as a gateway, the -interface modifier should
>>      be specified; the gateway given is the address of this host on the 
>> common
>>      network, indicating the interface to be used for transmission.
>> ---
>>
>>   iow, it derives the iface based on what iface the addr you give it
>>   lives on.
>>
>>> Even though a security association for the target address exists on
>>> enc0. Unfortunately, the device is not passing traffic to 192.168.0.57.
>>>
>>> I assume I need to add a route -- but is this even necessary?
>>   you will need to add a route to the other end of the tunnel
>>   so that traffic that originates on the local endpoint destined
>>   for an addr matching the remote addrspec (Destination in netstat
>>   -rnf encap) goes over the tunnel.
>>
>>   if traffic originates on the local host and a matching route 
>>   is found in the inet (or inet6, i suppose) table, that route
>>   is taken.  if you have a default route, that will catch it
>>   (probably undesired), so you need an inet route to make it
>>   match something more specific than the default route in the
>>   inet table.
>>
>>   traffic traversing the host (forwarded datagrams) will match
>>   the ipsec flows before they get looked up against the encap
>>   table (if this is not literally correct, it is the behaviour
>>   i've observed) and thus do not need a route.
> 
> I am talking about forwarded datagrams in this case. That's been my
> experience in the past, but it's not working right now.
> 
> I'm using a device to NAT traffic from an internal host to the private
> IP address the remote IPsec peer expects on my end.
> 
> When I ping from the internal host and do a tcpdump on the internal
> interface of the IPsec device, I see that the packets have been NATted
> correctly and I would expect them to be matched to the appropriate flow
> and passed through to the remote internal network -- yet I get
> "Destination host unreachable" from the IPsec device. I note that that
> is not the same as "No route to host", but I remain suspicious. I don't
> see any of my pings go out on enc0.
> 
> Just to eliminate the obvious -- the IPsec device is forwarding other
> traffic just fine, and I have other working tunnels.
> 
> -Stephen-
> 
> 
> !DSPAM:44bc7cd3248752146810636!

Reply via email to