logging reject sizes... very interesting results.

[Dan Merillat]

I've had reject size logging going for an hour and a half now,
and what I'm finding is pretty interesting.

(Filtered out some junk servers)

Incoming Feeds (INN):
Server                Connects  Offered   Taken Refused  Reject  %Accpt  Elapsed
sp25fe.nerdc.ufl.edu        87    94695   83087    5871    5737     87% 27:17:25
newsfeed-east.sprintlink    12    31646   21297    5690    4659     67% 59:45:06

TOTAL: 4                   127   126466  104506   11561   10399     82% 92:41:27

Server                      O-Size     Size  RejSize %Size 
sp25fe.nerdc.ufl.edu       10.0 GB   3.6 GB   6.4 GB   36% 
newsfeed-east.sprintlink    9.4 GB   1.5 GB   7.9 GB   16% 

TOTAL: 4                   19.4 GB   5.1 GB  14.3 GB   26% 
(Offer size/Accept Size/Reject size)

So 75% of usenet is spam by cleanfeed's definition?

It's actually worse then that, since the server has been up for 5 hours
accepting posts, while rejects in that time were counting as zero bytes (unlogged)

The patches are pretty trivial, but they make too many assumptions, so I've got
to do some cleanup.  Basically, in ARTlog, I assume if text != NULL, it's a
reject and tag on Data->SizeValue.  A few hours of playing with innreport and
it's set.

(I'm sure I've duplicated effort here, but I just could not find patches anywhere)

Any interest?

--Dan