How to understand spamassasin speed

[Glenn Steen]

On 04/02/2008, donald.dawson at bakerbotts.com
<donald.dawson at bakerbotts.com> wrote:
> > -----Original Message-----
> > From: mailscanner-bounces at lists.mailscanner.info
> > [mailto:mailscanner-bounces at lists.mailscanner.info] On Behalf
> > Of Marcello Anderlini
> > Sent: Monday, February 04, 2008 5:05 AM
> > To: 'MailScanner discussion'
> > Subject: How to understand spamassasin speed
> >
> >
> > Hello everybody,
> > Is there any way to test the speed of single spamassassin
> > test so to know
> > which is slowing my system ?
> >
> > I've done spamassassin -D --lint 2>/tmp/speed.txt but there
> > is now way to
> > understand how may time each process takes long ?
> >
> > Thanks for your help and sorry for my worst English
> >
> > Best regards
> >
> > Marcello
> >
> One of our new mail servers had been blocked by spamhaus and that was
> causing additional delays.
>
> I added the following to the spamassassin prefs file on our MX server
> and the processing time per message improved by 3 seconds.  Someone with
> more experience in this mail list can verify if this is correct, or too
> extensive in dropping RBL lookups.
>
> # 11/22/07 DLD - from ms listsever - stops spamhaus lookups
> score __RCVD_IN_ZEN 0.0
> score RCVD_IN_SBL 0.0
> score RCVD_IN_XBL 0.0
> score RCVD_IN_PBL 0.0
> score URIBL_SBL 0.0
>
> # 11/26/07 DLD Timeouts using ms debug
> score URIBL_RHS_DOB 0.0
> score DNS_FROM_DOB 0.0
>
> I ran:
>
> # MailScanner --debug --debug-sa 2>&1 | awk '{printf"%s %s\n",
> strftime("%T"), $0}' | tee /tmp/mstest.log
>
Granularity of seconds is a bit limited.... Use strftime("&t:%N") for
subsecond timings.
Granted, seconds will likely be enough to see any blatant problems.
Also, since we're debugging SA, it'd be better to just use
spamassassin -D -t < /path/to/message file | awk '{printf"%s
%s\n",strftime("%T:%N"), $0}' | tee /tmp/mstest.log
... IMO:)

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se