Just tracked done a bug in the serendipity_track_url() function:
In this function, a table entry is updated which the following conditions: scheme, host, port, path, query, day, entry_id. If affected_rows() is 0, then a new entry is inserted into the statistic table. This table has an unique constraint about three of the seven columns: host, day and entry_id.
So if i have 2 URLs in a blog entry, which have the same host but different pathnames, i get a SQL error about unique key violation, because it won't match the first update but will match the unique constraint.