It sounds weird to me, the 192.168.1.10 is on server eth0 network interface, and it will be on my local etc host file, will postfix still do NSlookup ?
since i try inject 100 msgs, there is not time out 5 secs then only start injecting for the 100 msgs, the 5 secs is the time for 100 msgs to inject to the queue. i have tested this example on the other server ( the cde.com ) which also running the DNS ( temporary for this test ) it took 1.5 secs for localhost and 3 secs for the eth0 interface on this server, for this server the time only increase 2 times, but on the high end server it tooks 10 times ? dns lookup can really have such a big impact ? i will try using dig command later to find out the ns lookup time between the server. On Tue, Jun 1, 2010 at 3:19 AM, Wietse Venema <[email protected]> wrote: > Teh Kim Chooi: > [ Charset ISO-8859-1 unsupported, converting... ] > > Hi guys, > > > > i recently just setup a high volume postfix server, still in testing mode > > before the server go for live, OS rhel 5.5 and postfix version 2.3.3 > > Which is no longer maintained. The last release was postfix-2.3.19 > in August 2009. > > > time smtp-source -s 100 -l 409600 -m 100 -c -f [email protected] -t > > [email protected] localhost:25 > > > > result is 0.5 second which looks fine, > > Here, your client IP address is 127.0.0.1 or ::1. > > > BUT when i use the server IP instead of localhost, my performance drop > > > > time smtp-source -s 100 -l 409600 -m 100 -c -f [email protected] -t > > [email protected] 192.168.1.10:25 > > > > result is 5 seconds, which almost 10 times more. > > Here your client IP address is 192.168.1.10. > > The Postfix SMTP server looks up the host name. Apparently this > times out after 5 seconds when looking up 192.168.1.10. > > Wietse >
