Re: Incoming feed statistics

At 01:59 PM 5/8/03 -0400, Jeffrey M. Vinocur wrote:
> >>> =ybegin part=2 line=128 size=10485760 name=[VE] Rurouni ...
>
>but it's apparently not getting caught by the yEnc filter I snarfed from
>Russ a while back (but looks like it should match just fine, unless my
>understanding of Perl is completely broken).

     Yeah, I have notice leaks from the yEnc filter I am using.  I am using 
one written for the old Cleanfeed.  It does have a bad flaw, it will only 
catch the yEnc articles if "=ybegin" is on the first line, first 
characters.  Aside from the flaw, I still see a high amount of yEnc 
binaries getting through that should not.  I have seen the yEnc articles 
being sent to us by peers I know are running different yEnc filters then we 
are (same filter you are using).  The yEnc filter is catching about 30,000 
articles a day, so I know its working and our volume is way down, 770Mb/day 
since I added the filter.  So I have kind of accepted the leaks of the 
filter for now but have hopes for improving it.

    I came to the conclusion when there is a high % of yEnc being put 
through the filter, the filter bogs down and can't keep up.  I have been 
using a solution I don't care for and was hoping it was a short term fix, 
by just removing the stealth newsgroups.  The filter I am using is:

                 if ($hdr{__BODY__} =~ /^(\s*\n)?\=ybegin/ ) {
                 return reject('yEnc post');
         }