|
Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com |
From: aio shin (aioshin
gmail.com)
Date: Thu Feb 21 2008 - 13:23:33 CST
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Maybe the load balancing part is right, but we have not yet tried that.
We suspect that amavis created a bottle neck because when we
temporarily disable amavis, and only postfix is working, no filtering
but have the rbl listing, the queue are cleared fast.
Now, we have tried increasing the Max server of amavis from 2 to 15
and also increase the amavis smtp unix process at master.conf to 15.
And it resolves the long queue hold time.
The only side effect is that it cpu usage increase as amavis works at maximum.
For now, that is the solution. it only relay the email to other
postfix servers, so in addition to reduce the load, we let some relay
domains to bypass spam and virus scanning.
Thanks for sharing your advice.
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]