> From:         Nick Gimbrone <NJG@CORNELLA>
> For the archiver to erase files it must have a R/W link. I assume the
> archiver is in a separate userid. This implies that LISTSERV has a
> R/O link and that to write on the disk it must re-link and access the
> disk. If it is releasing the disk when done writing on it then the
> directory is getting rewritten ok (which eliminates a few sources of
> potential problems). On the other hand, if the archiver is a task
> within LISTSERV, then this is a different problem... Can you clarify
> this?
 
The VMARCH MODULE runs in LISTSERV and talks through VMCF (maybe IUCV
by now) to the VMARCH virtual machine.  LISTSERV keeps R/W link and
access at all times.  Data passes through the intermachine facility
to get archived.  The VMARCH MODULE retains control while the file is
being passed into the VMARCH virtual machine.  Sometimes it gets slow,
too.  This is all done by the LSV$ARCH EXEC I wrote which only works
files created from lists keeping SEPARATE "notebooks", i.e. matching
the filespec "* %%-%%%%%".  It is customized somewhat to our local needs,
but anyone is welcome to use it.  I'll have to look up how to make
LISTSERV know the file can be had by anyone with GET.