NT 4.0: disappearing network protocol (TCP/IP)

[StefaniAkins]

This is the second time we've encountered this problem, and so far we have not found a solution that beats recasting the machine... please help.

A user had problems connecting to the network. Upon inspection of the network props, I discovered what I assume to be some kind of default IP address. We use DHCP, and sure enough, I received the error DHCP host could not be reached. Just to make sure, I tried renewing the address and failed.

So I checked the protocol settings. Imagine my surprise when the box came up blank. Of course, I attempted to reinstall TCP/IP, whereupon I got a message stating "TCP/IP version 4.0 already exists on the system. Please use [Update] to update the software." Well, since there isn't anything visible to be selected, nothing can be updated.

To continue: I found an article in the MS Knowledge Base that seemed relevant enough and required some scrubbing of the registry. After I made a backup, I once again tried to reinstall the protocol. This time, when I got to the point where files should be copied, the system locked up.

I'd be happy about any suggestions at this point.

 

[aalmeida]

First I think you should take a look on permition on you NT 4.0 you may not have permition to mess with Protocols, remember that administrator would have permitions unless specificaly denied (it could have happened by accident). AL Almeida
NT/DB Admin
"May all those that come behind us, find us faithfull"

 

[StefaniAkins]

That's true, but I'm a domain administrator, so no permission problems here. (guess I should've mentioned that)

 

[aalmeida]

On Login are there any scipst been loaded that could affect protocol Configurations?
One comon situation I have seen is that Admins expect everyone to be on DHCP but some Smart users see that the IP address even though is set to DHCP does not change so they configure it to be static for theyr purpose and when the admins load some new login scripts the TCP/IP stack and configuration gets confused.
Have you had any new scripts around the date the user pointed the problem? AL Almeida
NT/DB Admin
"May all those that come behind us, find us faithfull"

 

[Maruis]

There is a few thing you can do:

Try to release the IP via cmd.
Renew the IP via cmd.
If this does not work go to your DHCP server and force it to release that spesific machine.
Once you have done that repeat the first 2 steps.

The machine will now work.
I have found that sometimes for a unknown reason the DHCP server locks up on the IP address on one spesific machine. The rest of the network will have no problem. Most proberly that spesific machine will have a IP that is not in your DHCP range.

Hope this will help you.

 

[hibit]

Be careful do not jump to fast....

You said that the problem is that the DHCP host could not be contacted. This tells me that the client is sending the braodcast to attemp the lease. Look at network layer 2 issues first. Do you use cisco switches? or any switch that run Spanningtree?

If yes, are you able to do an ipconfig /renew_all after 5 min?

If you can get an address after 5min time check with your switch admin and if cisco have them go to the client port and issue portfast. This will let the port come active in 5sec rather than 90sec. Most of the time an MS box will timeout with spanningtree on and give up on the lease and send the error that you stated.

good luck