Vintners.net -- Understanding mailstats output
(Page last updated: 12-Feb-2003)

The percentage after each rejection item, represents the percentage of the total number of incoming emails bounced by this rejection rule. To a limited extent, this allows comparison of the effectiveness of each of the different rejection rules.

There are two caveats.

  1. The different rejection mechanisms are going to be hit in differing amounts. For example, the specific rules governing RFC-Ignorant listing are less likely to occur "in Nature" than an indiscriminate blocklist. Undoubtedly the size of the Osirusoft DB is much larger than that of RFC-Ignorant.
  2. Upon seeing these statistics, you'll want to compare the indiscriminate blocklists in order to determine which is the most effective. Your immediate reaction will undoubtedly be to consider the lists with the highest percentage of rejections to be the most effective. This is not necessarily a correct assumption.

    Although it may seem obvious, this does need to be called out:
    Bear in mind that the first blocklist consulted is the most likely to get a "hit". That list is most likely to have the highest numbers, as the test proceeds on to the next list only if the first didn't list this guy.

    Or, another way to look at it is; if all the lists have listed the creep we're looking up, only the first one gets the credit, as we stop looking as soon as we get a "hit".

    I believe that the order the lists appear in the .mc file dictates the order they are checked by the server at receive time -- however I have not looked at the sendmail code to confirm this.

    Now to really complicate matters, my mailstats script (which renders the report webpage) displays the blocklists sorted in alphabetical order, which does not match the order they appear in my .mc file.

  3. The Vintners.Net local blocklist is a list of creeps' IP addresses and domain names that I have manually typed in to the sendmail access file (details at sendmail.org) each time I have gotten a spam email since early 2002.

    The essential ingrediant statistics wise, is that this list comes first. A listing here overrides any other blocklist. Thus, this list will always have a leg up on the other lists -- not only is it "tailor made" to the spam my system gets, it's also checked first.

    Lastly, I started using this method before subscribing to any other blocklist. This means that the Vintners.Net local blocklist will undoubtedly get many of the hits that other lists would also get.


Copyright © 1997-2017 Vintners.Net (running on host vinifera)