My current settings are as follows:

THREADS 10
WAITFORMAIL     10000
WAITFORTHREADS  1500
WAITBETWEENTHREADS      100

WAITFORTHREADS will cause Decludeproc to wait the configured time if the
maximum number of threads configured are running before checking to see if
it can spawn another thread. Therefore, it is only used if Decludeproc has
reached the maximum number of threads. So, if your THREADS count is high,
this may have no effect.

WAITBETWEENTHREADS will cause Decludeproc to wait the configured time after
it spawns one thread before spawning another. So in theory, if you have
10000 that means it should wait 10 seconds before processing the next
message.

What I would like to see you do is put all three logs into DEBUG mode, let
some messages go through, change log level back, then extract from the logs
all the lines for a message and look at the time stamps to see how long it
is taking to process a message.

As far as I know you have to restart the Decludeproc service when changes
are made to the declude.cfg file. I am assuming that the Global.cfg file is
still read by each thread when launched. But since the Decludeproc service
uses the declude.cfg file to tell it how to run, it seems logical that is
only read at service start. Now, if I really wanted to know tonight I could
run Filemon and see what it says, but I have too much other stuff to do,
including something called sleep.

So for now, every time I make a change in the declude.cfg file or change the
Decludeproc.exe file I follow the order I have posted earlier to cycle the
services.

John T
eServices For You


> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:Declude.JunkMail-
> [EMAIL PROTECTED] On Behalf Of Kim Premuda
> Sent: Thursday, September 29, 2005 6:50 PM
> To: Declude JunkMail Forum
> Subject: [Declude.JunkMail] JunkMail 3.0.5 100% CPU Usage
> 
> IMail 8.05 HF3
> JunkMail 3.0.5.5
> 
> We are still seeing times when 'decludeproc.exe' will burst the CPU usage
to 100% for
> 20-30 seconds up to a minute or two. We've adjusted the WAITFORTHREADS and
> WAITBETWEENTHREADS to various values, and it doesn't seem to make any
> difference. The current values we are using are:
> 
> WAITFORTHREADS       10000
> WAITBETWEENTHREADS   10000
> 
> As an example, the 'work' folder will have several messages in process, so
you would
> expect to see the CPU usage go up for the first thread, then down to
minimal levels
> for 10 seconds, then back up again for each additional message thread
processed.
> But, what we see is the CPU usage hitting 100% and staying there while
processing
> the messages in the 'work' folder. If additional messages show up in the
'proc' folder
> and are transferred to the 'work' folder before the current THREAD run has
completed,
> they are immediately processed thereby keeping the CPU usage at 100%. When
> everything is eventually processed, the CPU usage returns to normal
levels. This only
> seems to be a problem under heavy load. Is anyone else experiencing this
behavior?
> 
> Also, does anyone know if the THREADS value can be reduced to a number
below the
> Declude recommended minimum of 5 (and not break anything)? My thinking
here was
> 
> Thanks!
> 
> 
> --
> Kim W. Premuda
> FastWave Internet Services
> San Diego, CA
> 
> --
> ---
> This E-mail came from the Declude.JunkMail mailing list.  To
> unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
> type "unsubscribe Declude.JunkMail".  The archives can be found
> at http://www.mail-archive.com.

---
This E-mail came from the Declude.JunkMail mailing list.  To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail".  The archives can be found
at http://www.mail-archive.com.

Reply via email to