From: Anthony Howe
Date: 2006-02-15 12:46:02 -0500
Subject: Re: 64bit support milter-limit
More information..: http://www.milter.info/#Support
Larry A Weidig wrote:
> Additionally after running this on the x86 box for almost a day
> we see a number of other suspect messages in the logs which I am hoping
> you might be able to offer some insight into:
> - Feb 15 10:55:00 sendmail: k1FGNhPR026531: Milter
> (milter-limit): write(Q) returned -1, expected 5: Broken pipe
> Feb 15 10:55:00 sendmail: k1FGNhPR026531: Milter (milter-limit):
> to error state
Either the milter died or the T= values in the INPUT_MAIL_FILTER() might
need tweeking w.r.t. to you server load:
> - Feb 15 10:08:00 sendmail: k1FG7jht027411: Milter
> (milter-limit): timeout before data read
> Feb 15 10:08:07 sendmail: k1FG7jht027411: Milter (milter-limit):
> to error state
Sounds like T= are too short.
Also are you using the default values for -g and -t options?
> - Feb 15 08:24:29 milter-limit: filterClose(): internal error,
> null workspace
Its a bit misleading this particular case. What appears to happen is
that the connection to aborted and closed before filterOpen() is called
and so when filterClose() is called, it reports that the workspace is
missing. I've seen this in some of my other milters and considered
disabling it in filterClose().
> - Feb 15 08:22:18 milter-limit: milter-limit: thread_create()
> failed: 12, try again
Now this one I've not seen before and looks like sendmail's libmilter
can't create a new thread. How many connections are you handling per
minute? Can you adjust the process thread limit higher?
Anthony C Howe http://www.snert.com/ +33 6 11 89 73 78
AIM: SirWumpus ICQ: 7116561 Skype: SirWumpus
Sendmail Anti-Spam Solutions
We Serve Your Server
Copyright 2009, 2012 by SnertSoft. All rights reserved.