On Wed, 2006-01-18 at 21:15 -0800, Wesley Peters wrote:
Please note that if you're running FreeBSD, moving to
a 'ram disk' won't help and will likely hurt. Make sure
your /var/spool filesystem has softupdates enabled.
You may want to try the 'noatime' option as well.
Out of curiosity, why
[EMAIL PROTECTED] wrote:
Even SpamAssassin shouldn't require locking unless the Berkeley DB
file is being written to. If it's only being read, then concurrent
access shouldn't be a problem.
Most bayesian analyses result in a write to add the new tokens and
update the token counts.
But I
[EMAIL PROTECTED] wrote:
3) Atomically rename database.db.new to database.db
Can you enlightne me on what Atomically rename means?
On Linux, type:
man 2 rename
From that man page:
If newpath already exists it will be atomically replaced (subject to a
few conditions - see ERRORS
[EMAIL PROTECTED] wrote on 01/19/2006 12:30:47
PM:
3) Atomically rename database.db.new to database.db
Can you enlightne me on what Atomically rename means? How is that
different than just a mv command. I can understand that a process that
opened the file under the old name is still going
On Thu, 2006-01-19 at 12:30, [EMAIL PROTECTED] wrote:
[EMAIL PROTECTED] wrote on 01/19/2006 12:30:47
PM:
3) Atomically rename database.db.new to database.db
Can you enlightne me on what Atomically rename means? How is that
different than just a mv command. I can understand that a
Hi David,
First question: do you have /var/spool/MIMEDefang on a RAMdisk? If not,
fix it now!
Next: It looks like you have 512MB of memory. You don't want to increase
MX_MAXIMUM much beyond around 20, or the server will start swapping.
I think moving /var/spool/MIMEDefang onto a RAMdisk
2006 19:42
To: mimedefang@lists.roaringpenguin.com
Subject: Re: [Mimedefang] BIG problems with mimedefang
Hi David,
First question: do you have /var/spool/MIMEDefang on a RAMdisk? If not,
fix it now!
Next: It looks like you have 512MB of memory. You don't want to increase
MX_MAXIMUM much
Lisa Casey wrote:
I think moving /var/spool/MIMEDefang onto a RAMdisk is a good idea. I've
not set up a RAMdisk before, but I've done some googling and think I can
get it done.
Question: I have 512 MG RAM on the system. Say I decrease MX_MAXIMUM to
15, how big should I make the RAMdisk?
Ben Kamen wrote:
Well, if you're doing enough traffic, you should make the ramdisk at
least as big as the max number of MD children you'd like to
run.. (I'd prolly shoot for 50% over that even.)
That's fairly conservative. You're unlikely to get all your MD slaves
processing maximally-sized
Hi,
I'm running Redhat 7.2 This computer functions as a Radius server (cistron
radius 1.6.7) and as a mail server (sendmail 8.12.6) which also runs
MIMEDefang 2.48 and SpamAssassin version 3.0.1 running on Perl version
5.8.5.
This setup has been working great up until this past weekend.
Lisa Casey wrote:
I discovered
that my max slaves was set to 10 and increased that to 30 hoping that
would solve the problem. It hasn't. Here's what I'm seeing. Things
will be going along ok, then suddenly the server load will shoot up
from 0.something to over 30! When that happens mail
Matthew.van.Eerde wrote:
Lisa Casey wrote:
Mem: 512900K av, 509840K used, 3060K free, 0K shrd, 1800K buff
Swap: 522072K av, 520104K used, 1968K free 5204K cached
PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND ...
11859 defang 10 0 30572 20M 1852 D 6.3 4.0 0:10 mimedefang.pl
11657
Lisa Casey wrote:
[classic symptoms of an overloaded system.]
First question: do you have /var/spool/MIMEDefang on a RAMdisk? If not,
fix it now!
Next: It looks like you have 512MB of memory. You don't want to increase
MX_MAXIMUM much beyond around 20, or the server will start swapping.
How
Hi,
I'm running Redhat 7.2 This computer functions as a Radius server (cistron
radius 1.6.7) and as a mail server (sendmail 8.12.6) which also runs
MIMEDefang 2.48 and SpamAssassin version 3.0.1 running on Perl version
5.8.5.
. . .
I have seen the same thing happen when the SA Bayes
Hi,
. . .
I have seen the same thing happen when the SA Bayes database gets
corrupt. Basically, SA hangs until MD times it out which causes the
whole slave to hang. Maybe someone here can give you a better
solution, but mine was to delete the database and relearn all my
corpus. SA rebuilds
From: Lisa Casey
Sent: Wednesday, January 18, 2006 12:02 PM
To: mimedefang@lists.roaringpenguin.com
Subject: [Mimedefang] BIG problems with mimedefang
Hi,
I'm running Redhat 7.2 This computer functions as a Radius
server (cistron
radius 1.6.7) and as a mail server (sendmail
Gary Funck wrote:
Lisa Casey wrote:
11859 defang 10 0 30572 20M 1852 D 6.3 4.0 0:10 mimedefang.pl
11657 defang 10 0 30172 9252 1820 D 3.0 1.8 0:07 mimedefang.pl
11652 defang 9 0 29184 8400 1832 D 2.9 1.6 0:07 mimedefang.pl
Try running strace on one of those busy slaves to see what
sort of
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of David
F. Skoll
Sent: Wednesday, January 18, 2006 2:03 PM
To: mimedefang@lists.roaringpenguin.com
Subject: Re: [Mimedefang] BIG problems with mimedefang
Gary Funck wrote:
Lisa Casey wrote
Gary Funck wrote:
If I recall correctly, 2/3 years ago, there was a particular version
of the Berkeley DB implementation that was bugging, esp. with respect
to locking (or lack thereof). It might've been in the Perl DB wrapper.
Could you be thinking of this bug?
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of David
F. Skoll
Sent: Wednesday, January 18, 2006 12:22 PM
To: mimedefang@lists.roaringpenguin.com
Subject: Re: [Mimedefang] BIG problems with mimedefang
Lisa Casey wrote:
[classic symptoms
David F. Skoll wrote:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=116192
That wasn't a locking bug. It was a weird bug whereby Berkeley DB
would, for no reason at all, sleep for one second whenever it needed
to allocate memory!
It's still present in Fedora Core 1, I believe.
21 matches
Mail list logo