Hey Guys,
I have stumbled upon something and wanted your advice.
I have a DAO class where I use mybatis3, something like this:

       function1() :
        try (SqlSession session = sqlSessionFactory.openSession(true)) {
            //select
            //update
            session.commit();
        }

      function2():
        try (SqlSession session = sqlSessionFactory.openSession(true)) {
            //update
            session.commit();
        }

      function3():
        try (SqlSession session = sqlSessionFactory.openSession(true)) {
            //insert
            session.commit();
        }

I am facing some deadlocks in postgres with this code under heavy load. Now I wrote some integration tests, and I was able to reproduce the deadlock fairly consistently. I do not really see why the deadlock should happen. I did note though in my code, I am opening session with autocommit true, so I figured I do not need to do an explicit commit. I made that adjustment and removed the session.commit(); lines from each of these functions. And now using the same test case, I am not able to see deadlock problem.

I tried several times and results are pretty consistent: With explicit commit, I do see deadlock issue in postgres. Without explicit commit, I don't see deadlock.

Question: Was/Is deadlock side effect of me doing autocommit session and simultaneously doing an explicit commit ? Can anyone explain this behavior to me?