While this question is surely off topic for LSTSRV-L, it's the only
place that I can think of where someone else may have seen the problem.
 
Environment: LISTSERV 1.8b, sendmail 8.6.12, Solaris 2.4
 
When LISTSERV starts up it opens a SMTP connection with sendmail.
sendmail responds by fork()ing a child process to manage that SMTP
connection.  LISTSERV then proceeds to hand message after message
to that child process which in turn fork()s other child processes
to actually deliver the mail.  As long as the inter-message time
period is less than sendmail's cached connection timeout the original
child process remains, receiving messages from LISTSERV and fork()ing
further offspring.
 
[log in to unmask] handles enough traffic that the cached
connection rarely, if ever, times out.
 
The problem I've encountered is that the sendmail process managing
the SMTP connection with LISTSERV get larger and larger.  A few days
ago it had grown to approximately 24MB and all of its offspring
being created were also 24MB.  This ultimately resulted in copious
quantities of "can not fork() no space" messages in /var/adm/messages.
 
Now for my question.  Has anyone else seen this behaviour?  Especially
on something other than Solaris 2.4?  I.e. is it unique to my system
or is it the nature of sendmail?
 
I suspect that if you are not running LISTSERV there's almost no chance
that a single SMTP connection to sendmail would receive enough traffic
to show the problem.
 
Glenn A. Malling <[log in to unmask]>
Syracuse University Computing Services   +1 (315) 443-4111
220 Machinery Hall  Syracuse, New York 13244-1260