DNS Requests Redirected to NDS

This is a repost of a message posted to a different group:

I would appreciate any suggestions to resolve this problem:

Here's the environment:
3 NetWare 4.11 Servers (SP9)
1 is the main File/Print Server
1 is dedicated to GroupWise 5.5 (GW is also patched to SP4)
1 BorderManager 2.0 (Proxy and Filter features turned off, but all
Internet traffic goes through this box)

2 NetWare 6 Servers
1 File/Print
1 DHCP

80+ Windows 98 SE Workstations, running Client v 3.31 or 3.32. Some of

the 80 are NT/2000 workstations runn Client 4.81 or 4.83. All pluggedinto a Cisco Catalyst switch.

Here's the problem:

Either between 7am and 8am, or between 11am and 1pm, the 4.11 Main
File/Print Server's Utilitzation goes up to 70% or more, where it's
normally 5% or less. Dirty cache buffers steadily increase into the
hundreds of thousands. The duration will normally be about 10 minutes,

then utilization will drop down to normal, and the dirty cache
bufffers will slowly decrease to 0. As soon as dirty cache buffers hit

0, normal service is restored. All the other Servers are normal during

this time. Of course, system response is very sluggish at these times,

causing a flood of calls to the IS Department.

Lanalyzer traces reveal something curious. Some workstations are
making DNS requests that somehow are being passed to NDS. For example:

"Req Resove Name (www.yahoo.com.LAW)" LAW is the name of the v4.11
Tree. The workstation sometimes makes this request to the 4.11
File/Print Server. The Server responds with "Rply Rqst Being
Processed, Conn=115" (Conn= varies, of course), but you never the seeexpected Reply with the address, or the "No Such Entry" response.
Sometimes, the DNS request will get handled by the NetWare 6 Server,
for example "Req Resolve Name (view.atdmt.com.FIRM)", FIRM being the
name of the v6 Tree. The NW6 Server will immediately respond with
"Rply Resolve Name; No Such Entry". Then, the workstation moves on and

make the proper DNS request: "Std. Req A vew.atdmt.com"

On the theory users are getting their non-work related Internet use
before the start of the day and and lunchtime, and that some glitch in

the Novell Client is causing these DNS requests to go awry, we've gone

to each workstation and deleted IP from the list of protocols. The
probem didn't go away; it just happens less frequently. I've also
reinstalled the NetWare Client at several workstations, and chosen the

IPX only option. Do I need to continue this for all workstations, or
is there something else I should be doing? Is the DNS/NDS problem thereal cause of the high utilization or just a side issue?

Thank you very much for any response.

Note:

Since the original posting of this message, we've made some network
changes, getting rid of 1 of the NetWare 6 servers, Novell DHCP and
Bordermanager, leaving just the 2 4.11 Servers and 1 NetWare 6 Server.

(I've thrown in the towel, and we're migrating to a Windows 2000 based

network) We also reinstalled all Novell clients with the IPX only
option, which seemed to cure the problem until this week. We checked
out each station again, and found about a dozen stations that we had
either missed, or which had client re-installed for whatever reason
with IP and IPX. So the Novell Client definitely seems to be the
cause. I got one reply from the other newsgroups I posted from someone

who thought there was an issue with winsock2. I suppose the problem
will cure itself when we complete the Windows 2000 migration, but whoknows; the IT budget might get cut and we might never do it, so any
ideas on how to fix the problem are appreciated.