On Fri, 26 Apr 2002 19:31:57 -0700 J C Lawrence <J> wrote: > Note: I run several multi-thousand subscriber lists and I have no lock > contention.
Minor stats: On a not very chunky system (PII-333) running a fairly untuned postfix can reliably sustain 1400 remote deliveries per minute (its actually a pretty stock Debian/testing install). The same system, which has a reasonable but not great IO chain (F/W SCSI III, single partition for /var which also holds /var/logs), can reliably spool (for remote delivery) in excess of 8,000 messages per minute (ie discrete SMTP transactions, not just RCPT TO's). I don't know if it can do more or what the actual top value is as I've not tried to find it and I've not had enough traffic to sustain loads at that level long enough. Note: Qrunner in 2.0 won't sustain that rate of spool entries as it is single threaded. I've created and sustained that level in previous tests running local custom qrunner variants against a hashed/split qfiles directory and thus plexing Postfix that way. You've talked previously of lists with low double digit numbers of subscribers, and of having what you insist are qrunner performance problems at that level. Again, qrunner is not your problem. An i486-33 with a bad IDE IO chain can easily, trivially, reliably sustain 20+ SMTP transactions (spool or remote delivery) per minute (I've done this). -- J C Lawrence ---------(*) Satan, oscillate my metallic sonatas. [EMAIL PROTECTED] He lived as a devil, eh? http://www.kanga.nu/~claw/ Evil is a name of a foeman, as I live. ------------------------------------------------------ Mailman-Users mailing list [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users Mailman FAQ: http://www.python.org/cgi-bin/faqw-mm.py
