[Mimedefang] This is fun... (temp failing)
Stefano McGhee
SMcGhee at ARCweb.com
Thu Sep 26 12:00:01 EDT 2002
Ashley,
I had similar problems with a P400 with 256 MB RAM. As I attempted
to increase thread timeouts and MAX/MIN slaves, I got poorer performance.
Duplicated messages, lots of temp failing and the like. It performed
better before tweaking. My solution was to move it to an 850Mhz with 512MB
ram. However, I would have though my previous machine would have been
adequate. I'm eager to see other people's ideas on this :)
Stefano
-----Original Message-----
From: Ashley M. Kirchner [mailto:ashley at pcraft.com]
Sent: Thursday, September 26, 2002 11:48 AM
To: MIMEDefang Mailing List
Subject: [Mimedefang] This is fun... (temp failing)
I have an old P200 with 224MB of RAM in it (odd number, it's an old
machine, okay?) running as our incoming mail gateway. So, sendmail, SA,
MD, File::Scan and ClamAv are all on it. Last night I changed the MIN/MAX
slaves from 3/5 to 5/10 respectively, setting the BUSY/IDLE timeouts to
60/300 respectively.
This morning I woke up, only to notice exactly one email was delivered
right after that change - and it was one I had pushed out of a quarantined
folder, so it basically came from localhost. After that Milter started
failing. This is the section of my maillog as soon as I restarted the
whole shebang:
==========
=== First, restart everything:
Sep 26 00:51:46 serpico mimedefang-multiplexor: started; minSlaves=5,
maxSlaves=10, maxRequests=100, maxIdleTime=300, busyTimeout=60,
clientTimeout=10 Sep 26 00:51:46 serpico mimedefang-multiplexor: Starting
slave 0 (pid 10915) (1 running): Bringing slaves up to minSlaves Sep 26
00:51:47 serpico sm-msp-queue[10930]: starting daemon (8.12.6):
queueing at 00:01:00 Sep 26 00:51:47 serpico sm-mta[10925]: starting daemon
(8.12.6): SMTP+queueing at 00:01:00 Sep 26 00:51:49 serpico
mimedefang-multiplexor: Starting slave 1 (pid 10935) (2 running): Bringing
slaves up to minSlaves Sep 26 00:51:52 serpico mimedefang-multiplexor:
Starting slave 2 (pid 10936) (3 running): Bringing slaves up to minSlaves
Sep 26 00:51:55 serpico mimedefang-multiplexor: Starting slave 3 (pid
10937) (4 running): Bringing slaves up to minSlaves Sep 26 00:51:58 serpico
mimedefang-multiplexor: Starting slave 4 (pid 10938) (5 running): Bringing
slaves up to minSlaves
=== Okay, everything started okay. Now let's push the message out:
Sep 26 00:52:04 serpico sendmail[10944]: g8Q6q4l6010944:
from=<bounce+OM_18122191683 at cmates.com>, size=12297, class=0, nrcpts=1,
msgid=<1494175047.1033021523976.JavaMail.cm_apps at rainadm02.sea2.cmates.com>
, relay=root at localhost Sep 26 00:52:04 serpico sm-mta[10945]:
g8Q6q49t010945: from=<bounce+OM_18122191683 at cmates.com>, size=12448,
class=0, nrcpts=1,
msgid=<1494175047.1033021523976.JavaMail.cm_apps at rainadm02.sea2.cmates.com>
, proto=ESMTP, daemon=MTA, relay=localhost [127.0.0.1] Sep 26 00:53:03
serpico sm-mta[10945]: g8Q6q49t010945: Milter change: header Content-Type:
from multipart/alternative;
boundary=635456838.1033021523976.JavaMail.cm_apps.rainadm02.sea2.cmates.com
to multipart/alternative;
boundary="635456838.1033021523976.JavaMail.cm_apps.rainadm02.sea2.cmates.co
m"
Sep 26 00:53:03 serpico sm-mta[10945]: g8Q6q49t010945: Milter change:
header MIME-Version: from 1.0 to 1.0 Sep 26 00:53:03 serpico
sm-mta[10945]: g8Q6q49t010945: Milter message: body replaced Sep 26
00:53:03 serpico sm-mta[10945]: g8Q6q49t010945: Milter add: header:
X-Scanned-By: MIMEDefang 2.21 (www . roaringpenguin . com / mimedefang) Sep
26 00:53:03 serpico sendmail[10944]: g8Q6q4l6010944:
to=<XXXXXXXX at pcraft.com>, ctladdr=<bounce+OM_18122191683 at cmates.com> (0/0),
delay=00:00:59, xdelay=00:00:59, mailer=relay, pri=30627,
relay=localhost.pcraft.com. [127.0.0.1], dsn=2.0.0, stat=Sent
(g8Q6q49t010945 Message accepted for delivery) Sep 26 00:53:08 serpico
sm-mta[10950]: g8Q6q49t010945: to=<XXXXXXXX at pcraft.com>, delay=00:01:04,
xdelay=00:00:05, mailer=esmtp, pri=30778, relay=mx1.pcraft.com.
[206.168.220.51], dsn=2.0.0, stat=Sent (g8Q6m4W08584 Message accepted:
There, doesn't that feel better?)
=== Message successfully pushed out, and remotely delivered. As soon
as that was done, I started getting these:
Sep 26 00:58:24 serpico sm-mta[10980]: g8Q6wO9t010980: Milter (mimedefang):
to error state Sep 26 00:58:24 serpico sm-mta[10980]: g8Q6wO9t010980:
Milter: initialization failed, temp failing commands Sep 26 00:59:47
serpico sm-mta[10983]: g8Q6xl9t010983: Milter (mimedefang): to error state
Sep 26 00:59:47 serpico sm-mta[10983]: g8Q6xl9t010983: Milter:
initialization failed, temp failing commands Sep 26 01:01:24 serpico
sm-mta[10998]: g8Q71O9t010998: Milter (mimedefang): to error state Sep 26
01:01:24 serpico sm-mta[10998]: g8Q71O9t010998: Milter: initialization
failed, temp failing commands Sep 26 01:02:11 serpico sm-mta[11004]:
g8Q72B9t011004: Milter (mimedefang): to error state
=== Now what?
Nothing else after this got delivered. For the rest of the night, this
is what kept filling the log file. So, this morning I reset my MIN/MAX to
3/8 (from last night's 5/10) and suddenly everything works again. Of the
several hundred emails that were queued up from all over the world to be
delivered to us, everything came through in a matter of about 20 minutes.
MD behaved just fine, informing remote machines when it ran out of slaves,
everything worked fine. All email has now been delivered and we're back to
our usual trickle of 30+ emails per hour.
What caused Milter to start complaining about MD and temp failing
everything?
--
H | "Life is the art of drawing without an eraser." - John Gardner
+--------------------------------------------------------------------
Ashley M. Kirchner <mailto:ashley at pcraft.com> . 303.442.6410 x130
Director of Internet Operations / SysAdmin . 800.441.3873 x130
Photo Craft Laboratories, Inc. . 3550 Arapahoe Ave, #6
http://www.pcraft.com ..... . . . Boulder, CO 80303, U.S.A.
_______________________________________________
MIMEDefang mailing list
MIMEDefang at lists.roaringpenguin.com
http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
More information about the MIMEDefang
mailing list