Thomas Hillson <[EMAIL PROTECTED]> writes: > The first involves qrunner, what happens is that a list gets locked and no > mail will go through it.
Tom, when I first moved my 1800-member list to mailman, it had upwards of 200 dead addresses. Those 200 dead addresses caused a lot of bounces, and for some reason the bounces seemed to reach critical mass, wedging the CPU and causing qrunner to abend. Qrunner was in cron, subsequent qrunner attempts started stacking up in the locks directory. First thing I did was to comment the qrunner cron job. I sent email to all 1800 addresses with a script: "I am getting ready to change the list, here's the status." (et cetera) I got back about 250 bounces, and I removed those addresses with bin/remove_members. Then I carefully deleted all the bounces from the qfiles directory, one *.msg/*.db pair at a time. I made sure no qrunner or other mailman processes were running, cleaned out all the locks and ran qrunner. It's been humming along at one-minute intervals since with no problems. List is ~1620 addresses, about 100 messages a day, none larger than 30kb. Morgan ------------------------------------------------------ Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
