Strange client problem 2

[dsimpaired]

Recently upgraded to NW6.5 and have been getting the clients upgraded since we removed ipx and have only ip now so the older ipx client machines are in need of help. (These are Windows 98 machines) During updates we have noticed that some of our Dell machines will not login to the server upon update to the newest client with all of the appropriate settings. They refuse to see the tree. Then, if you run winipcfg and release the ip and then renew the novell login suddenly pops up and it can see the network. Then after reboot it starts all over stating that it can't see the network until we renew the ip again.... Any suggestions? Am I missing something obvious? (wouldn't be the first time.)

 

[terry712]

are you using ciscos?

check the portfast enable on them
force the nic speeds

most importantly - check slp
#

 

[dsimpaired]

Yes I am using ciscos, manually set the speed across the board? (i.e 100mb for each port?)

Also, when you say check slp, What & where?

 

[ITsmyfault]

1 - portfast - helps w/DHCP issues. STP routine requires a 30-second period to look for loops etc. Portfast cuts this down so the hosts get addresses the first time..
2 - speed and duplex - try and hard code this on all hosts and all ports if possible. don't count on auto-neg to take care of things.. you may get lucky, you may not. hard core folks will insist you force 100-half as that is a clearly defined standard where 100-full is not (this is 3rd hand.. I can't sit and read RFC's all day! ) I force everything here to 100-Full and am happy.
3 - SLP - did you set it up? Did you point your clients to it? (client--->properties-->service location tab) this will speed the login process big time and also make mapping drives over a VPN (for ex) nice and easy.

From your initial desc, it sounds like you have stuff set up fine but DHCP is not happy.. I'd do the portfast settings first, then go from there fwiw.

 

[terry712]

best way to check slp from the client is type slpinfo /all

 

[dsimpaired]

Thanks to everyone, found an article that describes several things for cisco configs that could help the symptom. The line that I added to the ciscos was the following:

spanning-tree vlan 1 forward-time 4

"Type “spanning-tree vlan 1 forward-time 4” and press Enter. Cisco will NOT display any console messages after pressing Enter. This will set forward-time from a default of 15 seconds to 4 seconds, allowing quicker DHCP connections and prevention of timeouts from the cluster compute nodes"

There was also a command for SWITCHPORT HOST that would enable portfast but I didn't have to do it because once I changed the first setting the workstations immediately started logging in. Here is a link for the full article if anyone is interested

http://www.liniac.upenn.edu/software/xcat1.2/cisco2950-HOWTO.html

Thanks again everyone, as always you really come through in a pinch!