Sorry, forgot to CC all:
No the weight=1 issue is not yet resolved. In fact, I have been able to
determine that snf4sa is actually querying snfserver properly. I
removed the old plugin so only snf4sa is loaded by SA. I then tailed
the sniffer log and see items like this continuing to scroll by:
<s u='20090515130114' m='/tmp/snf4sa/dL5Q6vQZ9G' s='52' r='2266218'>
<m s='52' r='2266218' i='906' e='949' f='m'/>
<p s='0' t='44' l='65536' d='56'/>
<g o='0' i='67.23.34.175' t='u' c='0.936317' p='-0.0474465'
r='Normal'/>
</s>
Note the path to the temp file /tmp/snf4sa/....
That tells me that everything is working properly except the returning
of the score to SA.
I have tried running test messages through SA manually and the SNF4SA
headers get inserted properly, but I haven't yet run through a message
that sniffer identified as spam. I will attempt to get one of those and
run it through SA manually to see if SNF4SA returns the correct weight
when it identifies the spam.
I will also join the amavisd-new list and see if anyone there can shed
some light.
Dan Horne
TAIS
Director of Operations
www.taisweb.net
supp...@taisweb.net
828.252.TAIS (8247)
> -----Original Message-----
> From: Pete McNeil [mailto:madscient...@armresearch.com]
> Sent: Thursday, May 14, 2009 6:27 PM
> To: Alban Deniz
> Cc: Dan Horne
> Subject: Re: [sniffer] Re: SNF4SA - Message Sniffer Antispam Plugin
for
> SpamAssassin
>
> Alban Deniz wrote:
>
> <snip/>
> > > 1) I'll look at the SA3 and SNF4SA plugins to see if I can
determine the
> > > reason for the timeout, and a solution. Pete mentioned that one
major
> > > difference is that SNF4SA uses a TCP connection to communicate
with
> > > SNFServer, while SA3 uses SNFClient.
> >
> >
> > The only possibility I can think of is that the snf4sa plugin
doesn't
> > wait long enough when running under amavisd-new. The timeout in
snf4sa
> > is set to 1 second, which is long enough when snf4sa is run by the
> > spamassassin command line. It might not be long enough when running
> > under amavisd-new. I don't think this is the problem. However, if
you
> > don't mind trying a longer timeout, here's how to change it: Edit
> > snf4sa.pm, changing line 72 from
> >
> >
> > $self->{SNF_Timeout} = 1;
> >
> >
> > to
> >
> >
> > $self->{SNF_Timeout} = 10;
> >
> >
> > Of course, a 10 second delay to process an email is unacceptable;
this
> > would simply point us in the right direction. Please let me know if
> > can try this.
> Hey guys...
>
> The timeout used in the SNFClient is on the order of 30 seconds--- 10
to
> get a connection, 20 more to get an answer. When a system is busy it
can
> take a few seconds for other requests that have already started to be
> processed. The overall throughput is much higher than the individual
> message timeout may suggest.
>
> I recommend allowing at least 10 seconds -- though 30 might be more
> appropriate.
>
> Note also that I've seen SA itself take as long as 10-15 seconds to
> process a message (depending on conditions) and it is roughly nominal
to
> see it take 1 - 3 seconds per message in many configurations. SNF is
> usually much quicker -- but we can't make assumptions about what else
> may be happening on the system at any moment -- especially during
> start-up conditions where incoming messages might be queued elsewhere
> and ready to cause a rush.
>
> Also -- isn't it reasonable that if SNF4SA does timeout it should
> provide a 0 weight instead of 1 ??
>
> Is that issues resolved?
>
> Thanks for keeping me in the loop.
>
> _M
#############################################################
This message is sent to you because you are subscribed to
the mailing list <sniffer@sortmonster.com>.
To unsubscribe, E-mail to: <sniffer-...@sortmonster.com>
To switch to the DIGEST mode, E-mail to <sniffer-dig...@sortmonster.com>
To switch to the INDEX mode, E-mail to <sniffer-in...@sortmonster.com>
Send administrative queries to <sniffer-requ...@sortmonster.com>
