>As a non-backbone LISTSERV site with a small number of heavy network users,
>I think the local systems people would gladly give up some CPU to
>LISTSERV for the capability of doing global signoffs.  (I've been asked if it
>exists many times.)
>...
>If/when such a command is available, I bet the systems folks here would like
>to make it a routine part of a killing a userid--irregardless of whether the
>user was a "known" network user.  I would expect a lot of use of the command,
>so it is in our best interest to make it efficient.
>
>Steve Middlebrook
 
I know that I would make it a routine part of account deletion on my system.
I coordinate a large mailing list (TeX-L).  A large number of the subscribers
at the TAMVM1 peer are students on VAXes around campus.  The only word I can
think of to describe those accounts is "volatile".  Every time I send a
message to TeX-L, I receive about 10 rejection messages from accounts which
no longer exist on those machines.  I dutifully update the list, and then
three weeks later when I send another message, here they come again.  A GLOBAL
SIGNOFF command would be invaluable.
 
I also like the idea of sending the global signoff requests out in a batch at
night.  So what if users do it multiple times?  Send a quick ack for each one,
store the nodename and userid in some file, and then just before sending the
batch, do the equivalent of a Unix "sort | uniq" on the file to eliminate
duplicates.
 
---Glenn