I hope that you can read and understand my broken english ;-)

I have setup an site to site VPN with stand alone BM-Servers (every
BM-server has its own tree) at all ends (1Master, 3Slaves).



(I hope you can read this schema of the WAN ;-) In the 1st, 2nd and
3rd LAN are Novell-Servers (Lets name it "File-Server") who are in the
same tree (but own Partitions, of course). All BM-Servers are
NW6SP2, BM36SP1a. The DS-Version is the same on all File-Servers in
the tree.

DS-Repair shows -717 errors. To check the communication, i copy
ASCII-Files between the File-Servers with toolbox.nlm. The result: If
i copy files via VPN, all 1400Byte the file has 32Byte corrupt
Byte (filename date and time are correct). If i copy from an
w2k-workstation to a File-Server via VPN the file is OK. A copy from
w2k-workstation to a w2k-workstion ist OK, too. A packet-trace at the

received File-Server shows that two packets are received.

The first packed has 1486Byte (Protocol: NCP, Request "Write to a
file" (ncp=73)),
the second packed has 70Byte and waas protocol: TCP, the port numbers
of this traffic are varying,
then the receiving File-server sends a "reply to a file (ncp=73)"
And again a packe with 1486byte Request write to a file...
This loop continue a long as all bytes of the copied file are

The corruption of the file begin when the payload of the NCP paccket
ends (1400Byte), the rest of the payload is in the TCP packet, this
are 32byte.
I think that this are the bytes who are going wrong. It is correct
that part of the data flows in a NCP-Packet and the rest in an
TCP-Packet? Is it an VPN-Problem?

If i copy from an File-Server in 3.LAN to a other location, connectet
via frame-relay to the BM-Master (not shown in the schema above), then
there is no problem.

I'm really puzzled about this.

1000Thx, for any hints