Server losing its network connection 1

[esmithbda]

We have a Windows 2003 server that regularly loses its network connection. It was at first happening once a week, but now it is happening every night, so whatever the issue is, it is escalating.

The event log mentions that the server thinks another machine either has its netbios name, or its ip. The server has a fixed IP address that is outside of the internal DHCP scope, and nothing else uses its name (it is an odd enough name that it is nearly impossible for something to even accidentally step on it).

Once this happens, you have to physically restart the machine - restarting network services or TCP/IP on it resolves nothing.

The oddest part of it is that if you ping internal servers, or Google, it will usually time out - but sometimes it will make it through with a huge delay (2000+ms), and then back to timing out.

Any ideas on how to resolve this?

We are totally replacing this hardware in two weeks, so at this point we are just trying to get through this - but having to restart it everyday is getting very annoying.

 

[skialta]

Check your speed and duplex settings on the NIC. You may have a mis-match.

 

[esmithbda]

It is on auto-negotiate and is connected to a 10/100 network. It is connected to the same switch as our mailserver, which is the key server that it needs to be able to talk to, as well as the domain controller.

I could change it to force one of the suboptions for speed/duplex, but in the past I've seen more issues forcing it to a choice than leaving it on auto-negotiate (on other servers that I have tried that in the past with).

 

[skialta]

I'd check but you may have better success with setting to 100 full duplex (assuming your switch supports that). In previous companies I've worked at auto has always worked fine, but where I work today you have to hard-code your settings otherwise you'll have big issues. Perhaps it's the difference of the hardware (Cisco vs. Dell switches), I don't know...

 

[esmithbda]

Okay, I will give it a look and try that out - thanks for the tip.

 

[esmithbda]

A few days now have passed and I can say with certainty that this appears to have been the issue. I changed it from auto-negotiate to full duplex 100 and it immediately resolved it and we have had uptime ever since.

Thank you for pointing this out - as I had said, I had always heard to leave it on auto and have had issues otherwise.

 

[skialta]

Good to hear...Like I said, I've always been able to use auto except where I'm at now. Strange, but at least it's solved.