Len,
But I still feel there's something throttling amavis with the current
hardware. This machine groans on passing 900 - 1100 msgs/hour?
I think it can do a lot better.
The quickest overview on the sanity of amavisd/SpamAssassin is what
amavisd-nanny shows. Set $nanny_details_level to
Mark Martinec schrieb:
Is your amavisd older than 2.5.0 ???
If he had not updated meanwhile, yes:
Len Conrad schrieb:
FreeBSD 6.2
amavisd-new-2.4.3_1,1
p5-Mail-SpamAssassin-3.2.5
clamav-0.93.3
Marc
-
This SF.Net
Gerry Massat wrote:
Have you enabled the ram disk for amavis temporary files? That made a
huge difference on my system. For my system I have the following in
/etc/rc.d :
amavisd_ram=100M
For my 1gb system with 2 amavis processes, seems to do the trick for me
with no swapping. As I recall,
I've opened up amavis to 10 processes, not getting much help on the throughput.
This one-liner prints time, postfix-to-amavis delivery delay this msg, avg
delay all msgs
egrep 'relay=127.*:10024' /var/log/maillog | egrep -iv conn_ | awk '{c++;
a=substr($9,7); t+=a ; print $3, a,t/c}'
I'm
Hello there Len, it's been a long time!
Hello Eric,
I am running a 55W Quad Core Opteron with 4Gb of RAM with only 4 amavisd
processes and it runs like the wind
I think you should install more memory, I realize you put 1MB but really you
have 1GB yes? =)
As I said, when the machine was
On Mon, Sep 22, 2008 at 02:07:16PM -0500, Len Conrad wrote:
Hello there Len, it's been a long time!
Hello Eric,
I am running a 55W Quad Core Opteron with 4Gb of RAM with only 4 amavisd
processes and it runs like the wind
I think you should install more memory, I realize you put 1MB
Quoting Len Conrad [EMAIL PROTECTED]:
Hello there Len, it's been a long time!
Hello Eric,
I am running a 55W Quad Core Opteron with 4Gb of RAM with only 4 amavisd
processes and it runs like the wind
I think you should install more memory, I realize you put 1MB but really you
have 1GB
But I still feel there's something throttling amavis with the current
hardware. This machine groans on passing 900 - 1100 msgs/hour? I
think it can do a lot better.
You are definitely right; it seems to me there is still something
very wrong with your setup. Your messages should be taking
Len Conrad wrote:
But I still feel there's something throttling amavis with the current
hardware. This machine groans on passing 900 - 1100 msgs/hour? I
think it can do a lot better.
You need to find out where the bottleneck is. Most average
email messages should go through amavisd-new in a
Len Conrad wrote:
But I still feel there's something throttling amavis with the current
hardware. This machine groans on passing 900 - 1100 msgs/hour? I
think it can do a lot better.
You need to find out where the bottleneck is. Most average
email messages should go through amavisd-new in a
Len,
But I still feel there's something throttling amavis with the current
hardware. This machine groans on passing 900 - 1100 msgs/hour?
I think it can do a lot better.
The quickest overview on the sanity of amavisd/SpamAssassin is what
amavisd-nanny shows. Set $nanny_details_level to 2 in
FreeBSD 6.2
amavisd-new-2.4.3_1,1
p5-Mail-SpamAssassin-3.2.5
clamav-0.93.3
quad cpu 2 GHz
1 GB RAM , about 700 MB inactive or free
spamd about 5% weighted CPU
clamav about 10% wcpu
vscan (amavis), 2 instances taking 75% wcpu
cpu 0% idle
amavis max msg 1000*1024
postfix active queue (sending
vscan (amavis), 2 instances taking 75% wcpu
What is the iowait time on this process(es)?
-
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin
vscan (amavis), 2 instances taking 75% wcpu
What is the iowait time on this process(es)?
sorry, forgot to show that. iostat shows burtsy i/o with several seconds of no
i/o, so it's not disk congestion
iostat -c 30
tty ad0 cpu
tin tout KB/t tps MB/s us ni sy
Suggestions?
Yes, you should be running much more than 2 instances of amavisd.
was running 6, upped that to 10 about 1/2 ago. no change.
sockstat -4 | egrep vscan
vscanperl 36980 5 tcp4 127.0.0.1:10024 *:*
vscanperl 36979 5 tcp4 127.0.0.1:10024 *:*
On Thu, Sep 18, 2008 at 06:34:23PM -0500, Len Conrad wrote:
Suggestions?
Yes, you should be running much more than 2 instances of amavisd.
was running 6, upped that to 10 about 1/2 ago. no change.
sockstat -4 | egrep vscan
vscanperl 36980 5 tcp4 127.0.0.1:10024 *:*
so now I've got 10 vscan's democratically eating 100% CPU with avg cpu load of
10, 100+ MB RAM available.
spamd and clamav worker bees still doing nearly 0% wcpu.
Still doesn't seem right that amavis as an interface should be eating the
entire machine while the content-scanners basically are
so now I've got 10 vscan's democratically eating 100% CPU with avg cpu load
of 10, 100+ MB RAM available.
spamd and clamav worker bees still doing nearly 0% wcpu.
Still doesn't seem right that amavis as an interface should be eating the
entire machine while the content-scanners basically
On 9/18/08, Len Conrad [EMAIL PROTECTED] wrote:
so now I've got 10 vscan's democratically eating 100% CPU with avg cpu load
of 10, 100+ MB RAM available.
spamd and clamav worker bees still doing nearly 0% wcpu.
Still doesn't seem right that amavis as an interface should be eating the
Gary V wrote:
On 9/18/08, Len Conrad [EMAIL PROTECTED] wrote:
so now I've got 10 vscan's democratically eating 100% CPU with avg cpu load
of 10, 100+ MB RAM available.
spamd and clamav worker bees still doing nearly 0% wcpu.
Still doesn't seem right that amavis as an interface should
20 matches
Mail list logo