1st,, sorry for (almost) crossposting since I've asked a similar
question in sup/edir

Layout is a wan network with 7 offices, each beeing it's own partition
Root/Central HQ have one server holding replicas from the other
offices/partitions and each remote partition holds a replica for root.

So, now comes the "goof".
One office should be switched off, all users, gw objects, etc
succesfully removed. Then removed that remote server from the replica
ring of root. Last, but not least, I removed all other servers from
holding a replica of that remote partition. This was the obvious error

Trying to delete the server through C1, it states that the server must
be taken down first, OK = Done.
Problem now is, all other servers still think the server is up and
that partition has NO server having either a master or a r/w replica
of it. The "removed" server can not be seen in C1 but is still there
either through dsrepair or dsbrowse but can NOT be deleted.

Tip from other forum-group was to promote another sub-ref server to
master for that partition = Done. But partition is still empty in C1
and server can still not be seen or deleted properly.

So, can't get tree to successfull sync since one server is missing but
not dead according to ds.

checking external ref;
"Log file for server ".name.name" in tree "test"
External Reference Check
Start: Sunday, 2011 October 9 12.20.27 Local Time
ERROR -352 checking XREF ID: 00016F64, DN: CN=name...."

My options right now to get rid of this is either;
A= waiting for Novell sup to maybe have a way to simply delete the
server in question. ( cause I can't)

B= Turn the server back on again since it is NOT "reformated" or
anything and still available. If that's an option, then the partition
could maybe sync up again and the re-do the delete of the server with
another server holding a copy of the replica/partition.