Details

    • Type: Bug Bug
    • Status: Closed Closed
    • Priority: Major Major
    • Resolution: Fixed
    • Affects Version/s: JRuby 0.9.9, JRuby 1.0.0RC1, JRuby 1.0.0RC2
    • Fix Version/s: JRuby 1.1b1
    • Component/s: Core Classes/Modules
    • Labels:
      None
    • Number of attachments :
      2

      Description

      Once in a great while, it seems that some thread test in our suite runs forever. The trace appears to show the main thread yielding and many threads all stopped. I have not traced it any further, and it doesn't seem to be affecting any user applications.

      1. ant_test_thread_problem_dump.txt
        101 kB
        Charles Oliver Nutter
      2. openwferu_scheduler_test_dump.txt
        25 kB
        John Mettraux

        Activity

        Hide
        Charles Oliver Nutter added a comment -

        This isn't hurting anything real, so I'm gonna punt it to post 1.0. I have a few theories about what might be happening, it's just a matter of narrowing down which test/unit test is resulting in this and digging in.

        Show
        Charles Oliver Nutter added a comment - This isn't hurting anything real, so I'm gonna punt it to post 1.0. I have a few theories about what might be happening, it's just a matter of narrowing down which test/unit test is resulting in this and digging in.
        Hide
        John Mettraux added a comment -

        Hi Charles, Hi Thomas,

        hope you're enjoying Japan and Kamakura.

        So I've been trying to run OpenWFEru on top of JRuby these days and this bug here seems to be a total showstopper. Launching small business processes and playing with them is OK, but I've got a load test that stalls after 10 to 30 launches. I had the impression that a "sleep" call never returned.

        I also have a scheduler test (testing this component : http://openwferu.rubyforge.org/scheduler.html ) and half of the time, it blocks.

        Seems like you guys have enough test cases for it, I attached one of my dumps anyway (I have the test case at hand if necessary).

        Best regards,

        John

        Show
        John Mettraux added a comment - Hi Charles, Hi Thomas, hope you're enjoying Japan and Kamakura. So I've been trying to run OpenWFEru on top of JRuby these days and this bug here seems to be a total showstopper. Launching small business processes and playing with them is OK, but I've got a load test that stalls after 10 to 30 launches. I had the impression that a "sleep" call never returned. I also have a scheduler test (testing this component : http://openwferu.rubyforge.org/scheduler.html ) and half of the time, it blocks. Seems like you guys have enough test cases for it, I attached one of my dumps anyway (I have the test case at hand if necessary). Best regards, John
        Hide
        Charles Oliver Nutter added a comment -

        I haven't seen this one in a long time, but since it's thread-related I would like to get it truly resolved for 1.1. Unfortunately, we have no reliable test case. John Mettraux, do you have anything?

        Show
        Charles Oliver Nutter added a comment - I haven't seen this one in a long time, but since it's thread-related I would like to get it truly resolved for 1.1. Unfortunately, we have no reliable test case. John Mettraux, do you have anything?
        Hide
        John Mettraux added a comment -

        Hi Charles,

        please close it.

        It was fixed by Nicolas Modrzyk's patch about threads a while ago. ( http://jira.codehaus.org/browse/JRUBY-1157 )

        I thought this bug got closed at this point, sorry for the garbage.

        Congrats for the compiler.

        Show
        John Mettraux added a comment - Hi Charles, please close it. It was fixed by Nicolas Modrzyk's patch about threads a while ago. ( http://jira.codehaus.org/browse/JRUBY-1157 ) I thought this bug got closed at this point, sorry for the garbage. Congrats for the compiler.
        Hide
        Charles Oliver Nutter added a comment -

        Reported as fixed on 1.1. I've seen it still happen occasionally on 1.0, but it would be impossible to know which fixes to backport.

        Show
        Charles Oliver Nutter added a comment - Reported as fixed on 1.1. I've seen it still happen occasionally on 1.0, but it would be impossible to know which fixes to backport.

          People

          • Assignee:
            Charles Oliver Nutter
            Reporter:
            Charles Oliver Nutter
          • Votes:
            1 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved: