4 Replies Latest reply on Jun 7, 2004 8:04 AM by theimmortal

    PM Question

    theimmortal

      We are using JBoss-3.2.3 on Solaris and using the file-pm-service.xml as the persistence manager. We performed a stop on JBoss and there were approx. 91,000 messages in the tmp directory. Starting JBoss took approx. 1.5 hours. When the file-pm-service.xml was deployed, it seemed to remove all the messages from the tmp area. Then, the Message Cache started up and put them all back.

      I have searched the forums (for file-pm-service,pm, messages, etc), google, and such and have not had any hits on this issue. If anyone can illuminate my darkened mind it is appreciated.

        • 1. Re: PM Question
          genman


          The file-pm is not officially supported anymore and does not get called by default. I wonder if that was clear in the installation instructions? The best PM is the JDBC2 or 3, with your choice of JDBC supporting database. MySQL and Postgres are good choices, HSQLDB (the install default) doesn't have any problem with limited datasets, but YMMV.

          • 2. Re: PM Question
            theimmortal

            According to the instructions, HSQLDB is not a good choice either (my recent exp. with it agrees). I am working on the jdbc2-service for Informix (our DB of choice) as an alternative.

            I was really just looking for any explanation as to why the file-pm takes all the messages out of the queues, and the MessageCache puts them all right back (and possibly if there is a way to prevent that). I have googled and search all these forums with no luck.

            Anyway, just so I don't go chasing my tail again; does the same thing happen with the jdbc2 pms ? I am hoping that it does not remove all the messages and put them all back when you start JBoss ( 2 hour startup time != good ).

            • 3. Re: PM Question
              gorano
              • 4. Re: PM Question
                theimmortal

                Thanks Gorano! That explains things quite well.

                To sum up, this is a *feature* which will hopefully be fixed in a future release.