For each
newsgroup, the
news server maintains a list of information about all articles in the group, called the News Overview database. Each article in the group has exactly one entry in the list, which consists of the article's
Subject,
From,
Date,
Message-ID,
References,
Bytes,
Lines, and
Xref headers.
The database allows intelligent newsreaders to perform article threading in a more efficient manner. Instead of grabbing every single article in the newsgroup to perform threading (getting the Subject and Reference headers, etc.), newsreaders just grab the entire database for the newsgroup. This is much easier on the news server as well: sending 1 file to the client is a lot less work than sending several thousand. (This is the main reason why the News Overview database was created: in order to keep threading newsreaders from crushing news servers into the ground.)
Tin's author assumed that no individual record would contain more than 1024 characters. However, excessively crossposted articles will have huge Xref headers, and articles in long-lived threads will have huge References headers. Because the database contains these two headers, the records for such articles often do exceed 1024 characters. When tin processes records with more than 1024 characters, it unwittingly truncates its copy of the records to the first 1024 characters, and then complains about the records being bogus.
Thus, the "Bad Overview Record" error message occurs due to a bug in tin, and not because the database is actually corrupted. Just ignore the error message, or switch to another newsreader.
Editor's Note: from http://www.pitt.edu/~news/faq.html