OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
From: Alex Kramarov (alexincredimail.com)
Date: Thu Dec 27 2001 - 09:19:57 CST

  • Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

    > No, that conclusion is entirely wrong. You are not measuring MTA
    > performance.
    >
    > All you have shown is that mail leaves the machine at 15/second,
    > whether you use qmail or Postfix, using your specific configurations.
    >
    > In other words, the delivery rates aren't limited by the MTAs.
    >
    > Given sufficient network bandwidth and delivery process parallelism,
    > the delivery rates will eventually become limited by the MTA's disk
    > usage, and it is known that qmail hits the disk a bit harder than
    > Postfix does.

    well, since using a ramdisk for the queue is not possible due to the volume
    of the list, i will just have to write a wrapper, what i have picked up on
    the qmail list - using qmail-remote to try to send each email to the
    intended recipient, and feeding the email to the mta only if the delivery
    failed, using postfix as the mta - i like it alot for all the additional
    features. does anyone has such an implementation ?

    this is the general logic :

    maxchildren=1000;
    children=0;

    whenever SIGCHLD decrement children

    while (workdoto) {
    get_work();

    if (children == maxchildren ) wait();
    child=fork();

    # error conditions in parent
    if (child<0) {
    error, set maxchildrenlower, exit or retry;
    };

    # this is the child process
    if (child==0) {
    send mail using qmail-remote
    if !success
    send mail using the mta
    exit 0 # child finished
    };

    # this is the parent process
    if (child > 0 )
    children ++

    }

    # wait for children to finish
    while (children != 0) wait();
    exit

    Alex.

    -
    To unsubscribe, send mail to majordomopostfix.org with content
    (not subject): unsubscribe postfix-users