NAT Internet Connection Keeps Going Down

We've got network address translation running on our Windows 2003 server. We've got 7 workstation computers on the network.

NAT has been sharing our broadband ADSL internet connection for about a month and has been working perfectly. Each computer can access the internet from their workstation via the server.


Here's our problem:

At various times during the day (about 4 times a day) our internet access goes down. No workstations can view and web pages nor can the server.

To get it back online again we have to disable the internal network interface and then re-enable it. It only takes a couple of seconds to do so but it's beginning to be a real pain. There are no errors in Event Viewer and we've confirmed it's not a problem with our ISP.

Routing and remote access does not give any visible errors about anything being down or offline but it does seem as though the problem is being caused somewhere near to where NAT passes the internal network traffic onto the external internet interface.


Any help on this would be greatly appreciated.


Cheers.

 

I think that in this case you might be doing yourself a favor to get a router that does NAT and offload it from the server.

Otherwise with the lack of error information to use as a starting place, this one may be a bear to track down.

Might be worth trying a different NIC though.

 

we're going to call Microsoft Support. The client's not fussed that it costs £180 they just need it sorting. Even though they can get back online by disabling one of the nics and re-enabling it** the amount of times they're having to do it is annoying them somewhat.

** It doesn't matter which network card they disable - the internal one or the external one the same problem occurs with each.

with regard to the NAT enabled router, we've got the exact same setup our end that the client has got with R&RA running nat with the same ISP. we never have any downtime.

 

It doesn't matter which network card they disable - the internal one or the external one the same problem occurs with each

Be that as it may, you still might easily be dealing with a NIC problem on one or the other. If they lose internet but not intranet, probably the "external ". If both, probably the "internal ".

And yes, multi-homing the server and letting it to the ICS style job will work. However, it usually isn't as efficient as a good hardware solution (router /w NAT, firewall, DHCP, etc.).

If you are of a mind to do a little testing, try this:

- "External" NIC (with the internal disabled)
.. change the ip and netmask to match the internal network
.. figure out what is the largest packet you can use to ping and then ping one of the other internal NICs using the -t switch, the large packet size, and pipe the results to a text file. I'd let it go for about 30 minutes. Then check the test file for 'Request Timed Out' type errors. If there are none or very few and those few very scattered (so, not 4-5 in a row) then assume the NIC is fine.
- "Internal" NIC (with the external disabled)
.. same process

- If both of them check out good then set things up as before and have another PC on the network do the same ping test to the external gateway address.

Depending on which, if any, of the above show problems you will at least have a starting point for you or us or Microsoft to work from.