3 Replies Latest reply on Feb 14, 2012 10:29 AM by cgw

    JBoss 5.1 losing POST request in very specific circumstances

    cgw

      First off, my apologies if I'm posting in the wrong location.  I have spent an hour combing through messages and different search patterns looking for an answer.

       

      Configuration (Windows, Apache 443, JBoss 8010, PostgreSQL 5432)

      Libraries (Standard, Hibernate)

       

      I have an application that harvests data from a Common Access Card using a python script (mod_python) and formulates a username/password which is then placed in the SSL stack.  Subject request goes from Apache to JBoss on 8010 where our extended httpservelet's doPost() routine should take over.

       

      On a few user accounts only during a post (gets are confirmed to work) the request is null by the time it gets to our doPost().  Note that this design works perfectly for gets and 98% of all posts.

       

      tcpdump on localhost 8010 shows the data being transmitted from Apache to JBoss in transit thus affirming that JBoss is the culprit.  A Valve on the JBoss host that displays the request shows an empty Post for these particular posts that are failing.

       

      The python handler/technique was used in a Apache+Zope configuration without fail for years, the only real thing that changed was JBoss.

       

      Does anyone have a suggestion on where to begin or who to ask?  Should I include more detail?