Major Description
In RobotRulesParser, when Nutch creates a agent string from multiple agents, it combines agents from both 'http.agent.name' and 'http.robots.agents'. Along with that it appends a wildcard (ie. *) to it in the end. This is sent to crawler commons while parsing the rules. The wildcard gets matched first in robots file with (User-agent: *) if that comes before any other matching rule thus resulting in a allowed url being robots denied.
This bug was reported by @Markus Jelsma. The discussion over nutch-user can be found here:
http://mail-archives.apache.org/mod_mbox/nutch-user/201401.mbox/%3CCAFKhtFzBRpVv4MULSxw8RDRR_wbivOt%3DnhFX-w621BR8q%2BxVDQ%40mail.gmail.com%3E