[comp.mail.sendmail] Is sendmail 5.65 compatible with older binmail, ucbmail?

haynes@felix.ucsc.edu (99700000) (03/01/91)

We upgraded to sendmail 5.65 (non IDA) about last November.  Since then
I have seen a problem - maybe it was there earlier and I never noticed
it much, but it's definitely there now.  From time to time I will see
mail in the queue that is waiting for delivery to a local user, and that
local user will have a mail process.  Presumably sendmail and binmail
are waiting for the user to unlock the mail file so they can complete
delivery.  But these ucbmail processes may be a day or two old; and when
I kill the oldest one the whole set of hung mails and sendmails
collapses like a house of cards.  So I wonder if there is something
incompatible between the older ucbmail and the latest sendmail that
leads to these apparent deadlocks sometimes.

This is on VAXen, some running Original(TM) 4.3 BSD and some running
Mt Xinu More/BSD.

rickert@mp.cs.niu.edu (Neil Rickert) (03/01/91)

In article <12961@darkstar.ucsc.edu> haynes@felix.ucsc.edu (99700000) writes:
>
>We upgraded to sendmail 5.65 (non IDA) about last November.  Since then
>I have seen a problem - maybe it was there earlier and I never noticed
>it much, but it's definitely there now.  From time to time I will see
>mail in the queue that is waiting for delivery to a local user, and that
>local user will have a mail process.  Presumably sendmail and binmail
>are waiting for the user to unlock the mail file so they can complete

 Whatever your problem is, it is very unlikely to have anthing to do with
'sendmail'.  Since 'sendmail' does not deliver to mailboxes, it doesn't
ever lock them.  It totally relies on other commands, such as binmail, to
do the delivery.

 The only time that 'sendmail' delivers directly to mailboxes is when you
use a filename as an alias.  But even then 'sendmail' just appends the mail
to the end of the file without locking.  (The code comments imply that the
code has to be able to dispose of problems fast, such as dumping mail in
dead.letter when there are system problems).

-- 
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Science               <rickert@cs.niu.edu>
  Northern Illinois Univ.
  DeKalb, IL 60115                                   +1-815-753-6940