DHCP provided address needs manual renewal

[Shekinah]

I am running a network with two win2k server running DHCP. I only have 25 computers on my network so this is for fault tolerance.
DHCP1 has a scope of 10.10.10.15-98
DHCP2 has a scope of 10.10.10.103-250
My lease length is 10 days. The scope and server options on both servers are identical.

All clients run xp Pro with SP1 and all updates applied.

All clients are configured as DHCP clients.

All clients are on the same subnet.

Five of the clients will not maintain their connectivity to e-mail (hosted externally) or the web. Ipconfig/renew sometimes fixes the problem. Sometimes, to get the e-mail I have to combine this with Ipconfig/flushdns.

Here is what I don't understand. Why only on 5 clients? Sometimes I have to renew the IP several times a day. If the lease is 10 days, then the first time a renewal should be sought is after 5 days. Even then the address should be valid until the client panics at 87.5% of the lease and goes wherever it can get a valid address - i.e. my other dhcp server.

These are not limited to addresses from one scope - it does it with addresses from both servers.

Please save me so I have SOME hair left!

 

[Cervantes]

Hi Shekinah;
I had a similar problem with a client that would not (could not) renew it's address when it's lease came due. I'm afraid I don't know the nitty-gritty of why it worked, but releaseing the IP, changing the machine name, and rebooting so it leased a new IP with a new name, seems to have solved the problem. Maybe this will help.

Cheers,
Cerv

 

[GlenJohnson]

Five of the clients will not maintain their connectivity to e-mail (hosted externally) or the web

Same OS's? Anything in there log files?

Glen A. Johnson
If you're from Northern Illinois/Southern Wisconsin feel free to join the Tek-Tips in Chicago, Illinois Forum.
TTinChicago
Johnson Computers

 

[netometer]

Hi!
Can you gather some more info next time when you have that problem.
1.Run "IPCONFIG /ALL" and check whether you have IP from the "10.10.10.X" network or you lost it and the IP begins with the APIPA address 169.X.X.X.
2.If you have DHCP assigned IP can you try to ping it from the same machine.
3.If step 2 works fine - try to ping local IP addresses, then the Gateway then Internet IP addresses (like 198.133.219.25 , which is

http://www.cisco.com).

4.If step 3 does not work your problem is on layer 3 or below. Continue with changing the speed and the duplex of the card (I would suggest to start with 10Half). Next try to change the ports on the switch - I've had exactly the same problem with bad switch ports.
5.If step 3 works proceed with the name resolution troubleshooting.

My feeling is that the problem is with the NIC and switching the speed to 10Half instead of Auto will fix the problem. Three weaks ago I had a similar problem. i've installed 10 identical HP d530 desktops connected to the same Cisco 2900 switch. After a couple of days two of them lost connectivity and I had to do the procedures that you were describing. After doing this a couple of times I switched to 10Half and forgot about that problem.

Good luck!

http://www.netometer.net

 

[Shekinah]

Thanks everyone for your input. I know it takes time to do it. Glenn, as with most other forums I read with your input, your's has been most helpful here. I got interrupted when scanning the logs and missed that one thing in common seems to be inability to authenticate certai prosesses with AD. I am in the middle of making sure the affected machines are truly joined to the domain. I will then see if I can get through a day or two without having to renew the IP and let everyone know for future reference.

I am relatively new here but suspect that when the machines were joined to the domain that the netbios name may have been entered in a slot that wants a DNS name and DNS when it wanted netbios. Anyway, I will let you know.

Thanks again everyone.