Andrew Hodgson wrote:
I am currently using the default of Mailman sending out the message to the MTA (default Debian Exim on localhost) as one message with lots of recipients.
Is it best to fan out these deliveries into more than one message by changing SMTP_MAX_RCPTS and using more than one connection by changing SMTP_MAX_SESSIONS_PER_CONNECTION?
Lowering SMTP_MAX_RCPTS and/or limiting SMTP_MAX_SESSIONS_PER_CONNECTION will generally not improve performance.
All that is done by setting SMTP_MAX_SESSIONS_PER_CONNECTION to a non-zero value is to tell Mailman to close and reopen the SMTP commection to the MTA after that number of transactions. This is to accommodate MTAs that limit the number of transactions in a single session.
Likewise, lowering SMTP_MAX_RCPTS only tells Mailman to send more messages with fewer recipients per message. It doesn't add any parallelism to Mailman, although it may enable additional parallelism in the MTA.
With current Mailman, there is no way to add parallelism to the delivery of a single post from Mailman to the MTA.
Any other suggestions?
It has gone well so far, the only issue being initially blocked by Yahoo mail servers, however, I am dealing with this separately.
If you ultimately need to introduce parallelism in order to keep OutgoingRunner and Mailman's out/ queue from becoming backlogged with multiple messages, you can "slice" the out queue into segments processed by separate runners. See the section headed
##### # Qrunner defaults #####
in Defaults.py for information.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan