Re: Making spamassassin nice?
On (25/07/05 12:01), Jon Dowland wrote:
> On 7/25/05, David Baron <d_baron@012.net.il> wrote:
> > Thanks. /etc/default/spamassassin has limits on child processes and nicelevel
> > as well.
>
> Can anyone report on what sort of resources are needed to run
> spamassassin adequately? Since the upgrade to sarge, SA has totally
> killed our virtual machine. We have the lowest-tier bytemark VM (64MB
> mem). With a prefork limit of four processes and maximum nice, spamd
> still brings it to its knees. How does dspam compare in resources
> usage terms?
>
I'm not sure what quantative data would be of value but if you tell me
what output would be useful, I'd be happy to provide it.
We run spamassassin, clamav, exim4 as part of an IMAP server setup, on a
1Ghz PIII with 1Gb memory; it is also running apache/php/postgresql and
samba.
The child processes are set to a max of 5 and the nice level is the
default (ie. unset).
Snapshot of top:
top - 14:47:37 up 37 days, 4:03, 2 users, load average: 0.23, 0.14,
0.07
Tasks: 94 total, 1 running, 93 sleeping, 0 stopped, 0 zombie
Cpu(s): 15.6% us, 1.3% sy, 0.0% ni, 82.7% id, 0.3% wa, 0.0% hi,
0.0% si
Mem: 1036516k total, 1030416k used, 6100k free, 160416k buffers
Swap: 979924k total, 24k used, 979900k free, 571464k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
27046 Debian-e 15 0 40500 35m 6512 S 11.0 3.5 2:10.11 spamd
27037 root 16 0 40560 35m 6512 S 3.7 3.5 1:38.95 spamd
27035 Debian-e 15 0 40456 35m 6512 S 1.0 3.5 1:18.22 spamd
CPU usage seems to peak at < 20% but only occasionally; often spamd
doesn't even show in top.
Hope this is of some use. I suspect that the VM memory allocation is
the issue but what do I know? ;) I looked at bytemark hosting and it
looks pretty good but have yet to decide whether it's better to have
your own box hosted, if you're running resource hungry programs.
Regards
Clive
--
www.clivemenzies.co.uk ...
...strategies for business
Reply to: