It IS the same issue. I made the suggested change (i.e., adding nocanonify) and at the same time increased the number of Listserv mail queue processors - that significantly improved the situation but did not eliminate it. It has now become sufficiently annoying that I wanted to tackle it again. Now that I know it is a Sendmail problem rather than a Listserv issue, I will tackle it from that angle. On Tue, 8 Jan 2002 22:50:21 -0700, Ben Parker <[log in to unmask]> wrote: >On Tue, 8 Jan 2002 12:32:55 -0500, Gary Bannister <[log in to unmask]> wrote: > >>I would like to know how Listserv normally handles a posting to a domain >>that appears to no longer exist (i.e., the MX records for the domain can >>not be determined). With my current setup (1.8d under BSD/OS 4.2), when a >>posting is sent it is broken into envelopes of 100 addresses. If one of >>the addresses contains a domain for which the MX records can not be >>located, the message "Can not check MX records for recipient host <DOMAIN>" >>appears in the logfile and the posting will sit in the spool directory >>until the domain becomes reachable. > >Isn't this the same issue you raised last Sept and was answered by Valdis >in msg: >http://peach.ease.lsoft.com/scripts/wa.exe?A2=ind0109&L=lstsrv- l&F=&S=&P=8161 > >The short answer is that LISTSERV never delivers to end recipients by itself. >It always hands this off to an MTA (Mail Transfer Agent) such as sendmail. >Normally LISTSERV composes BSMTP 'packages' (in your case of 100 recipients) >and mails that 'package' to sendmail. LISTSERV has a very simple SMTP process >doing this that expects that the MTA will accept the entire 'package' without >question. Then when the MTA is attempting to deliver the actual mail, if it >finds a bad address, it is the job of the MTA to prepare an error reprot >message and deliver that back to LISTSERV. In short, the MTA must not attempt >DNS lookup verification on the domains at the time it accepts the mail from >LISTSERV. Valdis gave you some configuration recommendations for sendmail >that you reported did the work. Is that no longer the case?