General discussion

Locked

Netware Connections Dropping?

By gaztechs ·
-- We are running 2 Netware 5.1 sp4 servers.
-- Clients are mostly WINNT 4.0 sp6 with some W2K and WXP.
-- Novell Clients are mostly 4.83 with some 4.81. All are TCP/IP only, no IPX.
-- Switches are Cisco 2950's, dual homed to 2 Cisco 4006's. Servers are dual connected to each 4006 and clients are single connected to a 2950.
-- Clients login to the "nearest" server and attach to the other.
-- Here's the problem: Recently, clients began losing connection to only 1 of the Netware servers. It's always the same server and they are usually able to reconnect by launching the Novell client and logging in again. A pattern is yet to emerge. Clients that print frequently are the ones that report the problem most frequently but it has hit just about everyone at one time or another. Connections to other servers running Unix/Sybase, NT4.0/Exchange 5.5, NT4.0/Oracle, W2K and the other Novell server remain intact. We've had moderate success on some workstations by turning file caching off at the workstation and by moving the workstation to another switchport. The problem still persists but does not seem to be as severe. When an episode occurs, a Sniffer indicates both the workstation and the server begin re-transmission of un-acknowledged packets. The problem first appeared when we introduced a new W2K application server and removed a NT4.0 server that performed the same function. The NT server was a domain controller while the W2K server is simply a member server. Any help would be greatly appreciated.

This conversation is currently closed to new comments.

5 total posts (Page 1 of 1)  
| Thread display: Collapse - | Expand +

All Comments

Collapse -

by pgm554 In reply to Netware Connections Dropp ...

Among the things to look for:
1.Load scmd on the servers.There are issues with SLP that this seems to fix.
2.Make sure file caching is turned off at the file server level.
3.There are issues with the TCP stack on SP4.Make sure it is patched and up to date.
4.Do you have portfast turned on?If not ,turn it on.
The file caching thing at the workstations seems to point at that old Novell bugaboo :clent side file caching.This has been a pain for quite some time(since sp1)they always try to fix it ,but it has always been flaky.

Collapse -

by pgm554 In reply to

Servers are dual connected to each 4006 .
This definatly sounds like a port fast problem.

Are you load balancing at the switches or running failover?You need to watch this carefully.There is a 10 to 15 second delay with portfast enabled in acknowledging packets with portfast disabled(spanning tree)which is the default on the Cisco switches.

Collapse -

by pgm554 In reply to

Sorry ,some of the last post did not make sense.

Servers are dual connected to each 4006 .
This definatly sounds like a port fast problem.

Are you load balancing at the switches or running failover?You need to watch this carefully.There is a 10 to 15 second delay with portfast disabled in acknowledging packets.I remember a service pack for the NW clients that would send a ping every so often to make sure that the file servers were still there. Portfast disabled(spanning tree) is the default on the Cisco switches.

Also 4.83 sp2 (which works best with SP5)fixes a lot of flaky things.

Collapse -

by rglazer In reply to Netware Connections Dropp ...

You may be experiencing a few different problems.

(1) Check your NDS license make sure you have the correct amount, try to re-install it and point both servers to it.

(2) Check the server lan drivers for runts, crc errors etc.

(1)Perhaps you are experiencing NIC card driver issues with communicating with the Cisco switches. Some times the and NIC card speeds do not netogitate well, try forcing 100 FDPLX on server,switch port and workstations.

Collapse -

by Magpiper In reply to Netware Connections Dropp ...

This is a typical problem with the Broadcom LAN drivers. I recommend that you verify you have the latest drivers for your NIC.

Also verify that you duplex settings are correct for the switch and the NIC. Do not set to AutoDetect if you can help it. Try setting to HALF DUPLEX first and see if the problem goes away. Make sure the port on the switch (if it is a managed switch) matches what you change on the server's NIC.

Update to the latest TCP/IP patch from Novell

SET CLIENT ENABLED FILE CACHING = OFF

Verify SLP is working properly

Back to Networks Forum
5 total posts (Page 1 of 1)  

Related Discussions

Related Forums