[c-nsp] ADSL weirdness

Daniel D Jones ddjones at riddlemaster.org
Sun Aug 24 16:10:59 EDT 2008


This is driving me absolutely batty. I have an ADSL connection with a /29 
block of static IPs. I was originally using a BroadMax DSL modem. The modem 
works but locks up semi-regularly. Behind the modem, I have a 2651XM router. 
Tired of having to reboot the modem, I picked up an ADSL WIC for the router. 
After configuring everything, the router connects via ADSL and everything 
appears to be fine on the router. The problem is that I can not access some 
web pages. Hotmail.com and myspace.com are two that I know will not load but 
they aren't the only two. I can ping the web site IPs, at least those that 
answer ping. The page will start to load and then stall. Some pages will time 
out, others will simply say "loading" and never complete, even if left up 
over night. If I switch back to the BroadMax modem, I can load the same 
webpages without any issue.   I get the exact same behavior regardless of 
what browser I'm using, and on Windows and Linux, so it's unlikely to be any 
sort of host issue.

I've checked the web page IPs and there doesn't appear to be any pattern.  
They're certainly not all in a common subnet or anything.  The sites where 
I'm having  an issue do all seem to be more complex sites with lots of 
scripting.  I've tried to find out if the pages are doing anything weird, 
such as opening connections on unusual ports or transferring stuff using 
unusual protocols, but I haven't been able to identify anything.

I'm not at all certain that it's only web page traffic that has issues, but 
that's what I've noticed to this point. I run a mail server and a small 
mailing list, and I've gotten a couple of complaints of messages bouncing 
from one user but I believe that's his issue, not mine.  Mail appears to be 
flowing normally otherwise.

Here's the config I have on the router:

Interface ATM0/0
no ip address
no ip mroute-cache
no atm ilmi-keepalive
dsl operating-mode auto
!
interface ATM0/0.1 point-to-point
pvc 8/35
pppoe-client dial-pool-number 1

interface Dialer1
mtu 1492
ip address negotiated
ip nat outside
encapsulation ppp

dialer pool 1
ppp chap hostname USERNAME
ppp chap password 7 PASSWORD
ppp pap sent-username USERNAME password 7 PASSWORD

I'm running NAT for internal IPs but my servers have public IPs and the issue 
occurs regardless of whether I'm on a NAT'd machine or a server.

The internal networks runs on switches hanging off the fastethernet ports, 
which are the internal NAT interface. The servers are connected to a 16 port 
switch module in the router.  I was running some firewall rules but in an 
effort to solve this problem, I've removed all access lists other than the 
one liner which allows the private IPs into NAT.

Ideas, hints and suggestions all welcome.


More information about the cisco-nsp mailing list