As much background as I can manage:
Dell Servers, some 2800, some 2600. Some use Perc2.ham dated November
1,2004, others used the Pedge3.ham dated July 15, 2004. All have the latest
firmware (just compared against Dell site as of this date)

Netware 6.5sp3; in place upgrade on four servers, across the wire migration
to six others. Ten servers total. The four in place have traditional file
system SYS: volumes, all other volumes NSS.


Behavior described below is exhibited before any of the following updates
are applied--and remains when they are applied (in other words, they don't

NW65OS3a (to fix the cache memory allocation issue)

AFPTCP version 2.03.09, dated June 7, 2005, loaded with the "guest" switch.
Extended Mac namespace added to all volumes
AFP case-insensitivity added to all volumes

Clients: OS 9, 10.3.x, 10.4.x.

Issue: In OS X (Panther or Tiger) certain file copies cause the AFPTCP
process to hang on the server. There is no abend, at least the console
doesn't show one, and the log isn't written to. AFPSTOP will hang the
console. In most cases, the server itself can't even be restarted nicely,
not even with the debugger. The client performing the copy hangs as well.
Windows clients are unaffected (the few we have anyway), but additional
Mac clients can no longer log in. Already logged in Mac clients hang up as
well. Some may get the login window, be able to authenticate, and see the
volumes, but once they try to select a volume to mount, the workstation is

I can reproduce the issue by copying a large number of files from one
volume to another volume, or from a locally connected FireWire drive up to
a server. If I make the same copy (volume to volume or by FireWire drive)
using an OS 9 client there is no issue, and the copy is successful. It
seems to be only OS X causing the problem. Something in Novell's
implementation of AFP 3?

I have an open support call to Novell with this issue. It has been two
weeks, though, with no progress. I have sent them the afptcp.log file from
one of the servers, as well as a packet dump from the server and client
while the hang occurs. I have offered to send anything else they might
need, but they have not requested anything. Word is it is with Engineering
and Developement right now, and that this looks like an issue they have
fixed before, but I am fully updated, so they are looking into it. That's
great and all, but I have serious loads of users returning to school in the
next week, and this problem is only going to get worse. I have made
support aware of my urgency, but they seem to not be able to solve it.
Right now I have less than ten users on most of the servers, and we are
seeing the issue once or twice a day. Next week, that number will jump to
a couple of hundred per server. It isn't going to be pretty.

Does anyone have any insight into this issue?

Thanks a ton for anything I can try...