I'm running qmail-1.03 in a Compaq Proliant 6000 server with 4x200Mhz P-Pro
with 512Mb of RAM and an array of three U-SCSI 9Gb disks.
This server recives mails for my domain (is the MX), then it passes to
another smtp server that makes the local delivery to the mailboxes.
(specifing in the smtproute file the fixed route for the incoming mails).
These are some counts that I perfom with cat and grep in a highly hour :
cat /var/log/maillog | grep "Mar 28 18:3" | grep "new msg" | wc -l
181
cat /var/log/maillog | grep "Mar 28 18:3" | grep "starting delivery" | wc -l
114
cat /var/log/maillog | grep "Mar 28 18:3" | grep "info msg" | wc -l
181
cat /var/log/maillog | grep "Mar 28 18:3" | grep "end msg" | wc -l
99
cat /var/log/maillog | grep "Mar 28 18:3" | grep "success:" | wc -l
111
qmail-smtpd is invoqued by tcpserver, I start tcvserver instance with the
command "tcpserver 0 25 /var/qmail/bin/qmail-stmtpd &" in on of my init
scripts.
The Problem:
The problem is that i'm having long delays in the delivery of mails. In some
cases an email that arrives to the server can be stored for a couple of Hs.
in the /var/qmail/queue directory until is passed to the other host that
makes the local delivery.
I make a "find /var/qmail/queue/mess/. | wc -l" to see how many emails are
in the queue and this number continusly grow up.
Take in mind that this server don't make world deliverys, only accept
connections from the world and then pass the email to another host that is
in the same LAN.
I missed that this server is also running an apache web server with about
2000 hits/day, but I think this is not relevant.
The load average of CPU repoerted by the command "uptime" every time is
between 7 and 20.
The total of qmail preccesses are about 70 (ps -ef | grep "qmail" | wc -l)
Can any tell me if these values are OK for this hardware ?
Tnx.
Ricardo D. Albano
[EMAIL PROTECTED]
