Network Lock-Up with Dual NICs 1

[KingCanute]

Our office server (MS Small Business Server) has dual NICs, one for the internal LAN, the other to connect to the internet via an ADSL router/modem.

All generally works fine, but occasionally (roughly once per week, but no deterministic pattern), the network connection between the server and the ADSL router locks-up: the server cannot even ping the router and thus all outgoing internet access is disabled and (worse) all incoming SMTP email and VPN access is inhibited.

A successful work-around is merely to Disable the particular network connection (from the Network Connections screen), and then re-enable it. However, having to do this manually is a pain, and requires someone to be in the office (and we have people working in quite distant places around the world that need reliable access outside these hours).

First question is: Anybody seen this sort of problem before and can suggest possible causes/fixes?

But secondly: Is there a way of doing the Disable/Re-enable trick from a script (i.e. is there a command line equivalent)? If so, I can set an alert on the Health Monitor to automatically run such a script if a ping to the Router address repeatedly fails.

Thanks in anticipation,

Mike W

 

[MattWray]

Do you have a Gateway defined on your internal NIC, the one connected to the LAN? I have seen similar problems with a gateway on your internal NIC... Not sure about the scripting question...

Thanks,

Matt Wray
MCSE, MCSA, MCP, CCNA

http://www.sss-steel.com

 

[KingCanute]

Matt,

No, there is no gateway defined on the internal NIC, just a static IP address in the private (internal) range.

The DNS server address of the internal NIC is set to the same private static IP address, with the alternate DNS address set to the localhost address (127.0.0.1).

Cheers,

Mike

 

[excalibur78]

I have this same issue. I'm in need of a fix as well.

 

[justcurious]

What's new here? Has this always been a problem or did it start happening out of the blue? Think back to when it first started happening and try to identify even a simple change that may have occurred around that time.

Possibly the problem originates with the router and not your server. Next time this happens, cycle the power on the router to see if it corrects the situation.

When was the last time you updated the firmware on the router?

Check your duplex settings for the NIC in your network properties. Is it configured as recommended by the router/NIC manufacturer?

Is the server the only device connected on the WAN side? If there are other machines on that side, have you checked their host names and IP addresses for conflicts?

This may be a silly question, but have you checked your event logs?

 

[KingCanute]

Thanks for the response, justcurious, I really appreciate the input.

All excellent ideas, but the problem seems to be a bit more complex than this. In answer to your questions:

This has been an intermittent problem for us since we first installed the server (2 years ago) and has persisted through a change of NIC, a change of office (including new phone line, new ADSL connection and several changes of ADSL router, including a change in model and manufacturer).

Haven't updated the firmware on the router recently but, given the characteristics of the problem (as above), I'm pretty convinced that it's an issue with the operating system and/or the NIC device driver. We've kept the O/S up-to-date with latest service packs etc.

I'm waiting for a good chance to install a different NIC, but down-time is quite difficult to arrange on this critical component.

The NIC is set to Auto-Sense for duplex. I'm pretty confident that this is correct for the latest router, but I'll check.

There are no IP address conflicts with other devices connected to the router (there are usually none, anyway).

One thing that might be relevant, the problem occurs most often out-of-hours, particularly on the weekend. I thought that it was thus perhaps due to inactivity on the network link, so added a regular ping test (in Health Monitor) to keep the link from getting bored, but it hasn't made a noticeable difference.

Lastly, the event logs haven't shed any light on the matter.

Cheers again, any further advice gladly received,

Mike

 

[cfrishman]

First thing, if your nic drivers are the original from your install two years ago, update them. Next, find out if your switch is set to 100 mb full duplex. If so, get the nic off of auto sensing... W2k works better with switches when set to optimum duplex that can be run.

 

[KingCanute]

Thanks cfrishman, good suggestions and well worth a try.

Both the router and NIC were set to Auto-Sense, so I've forced them both down to 10 Base T duplex - it's only an ADSL link out of the router, so there's no need to make it any faster.

Also updated the NIC driver, so I'll let you know if we still get a problem (it can be several weeks between occurrences anyway).

Mike

 

[schnook9]

Hiya KingCanute,

We have noticed the EXACT same issue on our servers. This happens on not just one, but 4 servers. It tends to happen, just like you said during off-hours. I'm curious...
could you post up some more config info on your system? Maybe we can establish a pattern here and thereby find the cause of the trouble. We have had the issues with a 3com 3c905c nic, as well as an Intel pro 10/100 nic. We see this on dell poweredge 4400 and 1650. We are leaning towards it being an OS issue, but have not discounted the hardware completely. We also are running w2k fully patched, and I wonder if maybe one of the patches is causing an issue? just some thoughts.
Glad to know we're not the only one in this boat