In article <9A474DC3EA70D31194E600508B139CE050D951 AT lmig-msg-26.lmig DOT 
com>, Cris
Robinson <Cris.Robinson AT LIBERTYMUTUAL DOT COM> says:
>
>Does anyone know what would cause the database to deadlock while deleting a
>filespace if NO other processes were running?

Only a person with access to the source code or internals doc can
really answer the question. Which I am not. But I always quessed that
during the delete filespace process the code had to lock some
object in the data base that was shared by more then one filespace.

This might be something at the node level for nodes with more then
one filespace. Or maybe the records for multiple filespaces are stored
on shared data base pages.

Have you tried deleting only 3 filespaces at a time, I believe that this
should be free of the deadlock errors.


PS Before anyone falls on the floor laughing at the thought of the
   access to the objects in the first
   paragrah and adsm/tsm. just think, Microsoft
   supported the writing of a book titled "Inside Windows Nt" twice.
   and sun just announced free access to solaris source code.
       see WWW.sun.com