On Oct 20, 2011, at 11:30 AM, Wauford, Melissa wrote:

> On Oct 20, 2011, at 9:13 AM, Jacob Haller wrote:
>> On Oct 19, 2011, at 7:47 PM, Wauford, Melissa wrote:
>>> On Oct 19, 2011, at 6:32 PM, Jacob Haller wrote:
>>> 
>>>> Generally how LISTSERV delivers mail is by creating .mail files in ~listserv/spool/.  Then the SMTP workers (if SMTP_FORWARD_1 is configured) or the main LISTSERV process (if not) read the >>>.mail >files and deliver them to the outgoing mail server.
>>> 
>>>> It sounds like LISTSERV was creating a separate .mail file for each recipient, which happens for mail merge messages (since each recipient gets a slightly different copy of the message), or if >>>MAXBSMTP is set to 1 in LISTSERV's configuration, or under a couple of other sort of specialized situations.
>>> 
>>>> To prevent the sort of situation you describe, I would probably impose a limit on the size of messages, either using 'SizeLim=' settings on individual mailing lists or by setting a global limit via >>>FILEMAXL.
>>> 
>>> Hmm. I'm pretty sure this wasn't a mail merge. I don't think we ever turned on embedded mail merge (although perhaps it's on by default), and the list definitely doesn't have a mail-merge keyword. The log shows a standard mail message being processed. The only difference in the message files (I did a diff) was the "to" line (or whatever it is in pure SMTP). They looked like:
>>> 
>>> owner-listname*person**1SMARKETING*[log in to unmask]
>>> 
>>> Which is why I assumed the problem might be with the auto-probe setting.
> 
>> OK, that makes sense.  What is the mailing list's configuration?
> 
> Auto-Delete= Yes,Full-Auto,Delay(3),Max(10),Probe(1)
[…]
> Since the list is used maybe four times a year, the delay and max settings are obviously insane ( I didn't set up the list). I'm guessing that if the probe setting is really what's causing the individual messages in the spool, I probably need to bump up the value in the probe setting to at least 4. But maybe I just need to remove the auto-delete keyword and let the list owner deal with the bounces.

Yes, Probe(1) means every subscriber is probed every day there's a mailing, so that's why you saw what you saw.  For a list that is posted to so infrequently I would probably just remove subscribers on the first bounce, and use "Delay(1),Max(0)".  Regarding what Probe() number makes sense, Probe(4) would mean 1/4 of the subscribers would be probed, which is fine if you have the resources, but if it's a large list I might use something like Probe(10) or Probe(20) to probe 1/10 or 1/20 of the subscribers each mailing.  It'll take longer to get rid of some of the bad addresses, but it won't be as likely to kill the server whenever there's a big mailing.

Anyway, my two cents' worth!

Thanks,
jwgh
############################

To unsubscribe from the LSTSRV-L list:
write to: mailto:[log in to unmask]
or click the following link:
http://peach.ease.lsoft.com/scripts/wa.exe?SUBED1=LSTSRV-L&A=1