On Saturday, January 29, 2005, 9:15:23 PM, Glenn wrote: GR> This is question is a little off subject, but do you have any GR> recommendations for Imail queue manager settings? We are running Sniffer GR> with declude 1.82 under Imail 8.15 and the server seems to bog down GR> sometimes.
It is likely that your system is being pushed to it's limits. While I don't have any recommendations for queue manager settings I can recommend the following to help ease performance issues (these things work here in the lab anyway... ymmv) * In Declude AV (if you use it) use: AVAFTERJM PRESCAN ON * Be sure you're using a persistent instance of Sniffer: http://www.sortmonster.com/MessageSniffer/Help/PersistentHelp.html * It often helps to run a DNS resolver on your MTA and use the loopback address (127.0.0.1) to attach to it. I recommend BIND. It sounds counterintuitive at first - since this is yet another program running on the server, however in practice I find that things go much faster this way... theoretically because: -- Using the loopback address _may_ allow your system to skip some of the network stack - particularly NIC drivers etc thus eliminating some hardware and network related system loads and resource contention. -- Using the DNS server on your MTA eliminates all network delays / queueing. * Have plenty of RAM in the system. * Be sure your storage system is regularly defragmented and that you have sufficient free space. NTFS "REALLY HATES" a crowded and/or fragmented drive. * Be sure your spool is on it's own physical drive(s) if possible. If you haven't done this already, you'd be surprised how much performance a cheap, fast hard drive can add to your MTA when dedicated to your spool. The spool is constantly being read and written for what amount to temporary files. In addition, files from the spool are frequently going to be rewritten in your mailbox directories. If both of these operations are on the same physical device then you are guaranteed to impose the drive's seek time (and related OS queue related delays) to the processing of each message because the two operations will consistently compete with each other for the physical position of the drive head. Also, this particular mix of activity can effectively defeat any caching the drive may have by exceeding it's capacity and causing frequent cache flushing. Separating these operations on different physical drives results in far less drive head movement and a dramatic increase in the effectiveness of caching mechanisms in each of the separate drive systems. ___ I'm sure there are many additional opinions and hints floating around on this list both in general and also specifically for Imail/Declude installations. Hope this helps, _M This E-Mail came from the Message Sniffer mailing list. For information and (un)subscription instructions go to http://www.sortmonster.com/MessageSniffer/Help/Help.html
