Hi. I've been experimenting with large lists to try to determine the capacity of our current list server. (These one-time messages are not time-dependant, they just have to go out.) The first list I tried contained 41,000 addresses, all sorted and going to a single domain. I noticed that majordomo created only one entry in the mail queue, with all 41,000 addresses as recipients. Given that limited understanding I have of how sendmail works, it seems that perhaps this could be more efficient. I can't seem to find any parameter in majordomo.cf to address this. Does anyone have a way to do this? I'd like to break it up into 1000 - 5000 K chunks of addresses to see if it makes the mailing process more efficient.
Specifics:
Linux running on a Pentium with 48 M of RAM.
Sendmail 8.6.9
Majordomo 1.93
Perl 4.036
The way I understand it, the mail goes into the sendmail queue from majordomo, then we send out queue runners every ten minutes. I believe if it was broken up into smaller chunks, more than one sendmail process could be working on the list. I noticed that sendmail starts at the top of the list of 41,000 recipients (starts with the "A"s), and forwards the message with a certain number of addresses at a time to our more-powerful mailers that service the user population, then the recipient list just contains the B-Z addresses (approximately--really the chunks weren't that big). It took about six hours to process this particular list, at a good percent of CPU utilization, and I'd like to optimize majordomo before trying larger lists. We have a limited number of these large mailings that I can use for gathering data and tuning performance.
Any suggestions would be very welcome.
Bonnie Scott
Prodigy Services Company
Follow-Ups:
|
|
