We are running Netware 6 SP5 with NFA and Traditional volumes. On a Mac
os 10.4x and above, copying from volume to volume on the same server
fails. We have tried this on 10 servers with the same results. We have
differnt Mac hardware and different ethernet switch hardware, same
results. Copying to/from local drive works good. Copying from one
server to another works good. No errors were detected on the ethernet
switch ports.

Using Ethereal, I was able to capture the conversation between Mac and
Netware. I'm seeing 2 different scenarios;

1. Data is being read from the server to the Mac. A segment of data
appears to have been lost. The sequence numbers get out of wack. The
next expected sequence number is 17531477. The actual next sequence
number is 17534217. The sever sends some acknowledgements for data that
was written previously. The server retransmits the last segment 7
times. All the while, the Mac is not acknowledging the server. The Mac
sends a FPGetVolParms request and 4 DSI Request Tickle. The server
acknoledges all of the Request Tickle packets. After 15 seconds, the Mac
acknowledges the segment that was retransmitted 7 times with FIN set.
The server replys with RST set.

2. Data is being read from the server to the Mac. The Mac sends an
acknowledgement for received data. The Mac continues to send the same
acknowledgement 93 times. After the 9th time in only 5 microseconds, the
server has acknowledged this packet and continues to send data. The Mac
never acknoledges the new data, it keeps sending the same ack.
Eventually the sever retransmits some data and then resets the connection.

Any information would be appreciated!