Details

    • Type: Task Task
    • Status: Closed
    • Priority: Major Major
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 4.0-BETA, Trunk
    • Component/s: general/test
    • Labels:
      None
    • Lucene Fields:
      New

      Description

      common.test:
      [junit4:junit4] <JUnit4> says aloha! Master seed: 4CF29391D095CE45
      [junit4:junit4] Executing 2 suites with 2 JVMs.
      [junit4:junit4] Suite: org.apache.lucene.search.join.TestBlockJoin
      [junit4:junit4] Completed on J1 in 2.48s, 7 tests
      [junit4:junit4]
      [junit4:junit4] HEARTBEAT J0: 2012-07-11T14:27:14, no events in: 68.2s, approx. at: TestJoinUtil.testMultiValueRandomJoin
      [junit4:junit4] HEARTBEAT J0: 2012-07-11T14:28:14, no events in: 128s, approx. at: TestJoinUtil.testMultiValueRandomJoin
      [junit4:junit4] Suite: org.apache.lucene.search.join.TestJoinUtil

        Activity

        Hide
        Robert Muir added a comment -

        I think the one hour might be related to clover. Lets give it some time and see.

        Because if thats all it is, we could use the test multiplier here (clover uses 1x), so that we get
        more test coverage in nightly and hourly builds, without disrupting things. I just wanted to
        see if we can stabilize the test for now

        Show
        Robert Muir added a comment - I think the one hour might be related to clover. Lets give it some time and see. Because if thats all it is, we could use the test multiplier here (clover uses 1x), so that we get more test coverage in nightly and hourly builds, without disrupting things. I just wanted to see if we can stabilize the test for now
        Hide
        Martijn van Groningen added a comment - - edited

        Yeah one hour is way too long... even for the nightly build. Thanks Robert for fixing this!

        Show
        Martijn van Groningen added a comment - - edited Yeah one hour is way too long... even for the nightly build. Thanks Robert for fixing this!
        Hide
        Chris Male added a comment -

        Awesome +1

        Show
        Chris Male added a comment - Awesome +1
        Hide
        Robert Muir added a comment -

        unpatched: 62.87s, 14.08s, 6.05s, 41.36s, 3.90s, 8.14s, 12.50s, 16.09s, 7.02s, 7.18s, 125.19s, 37.17s
        patched: 3.03s, 3.76s, 3.22s, 4.20s, 2.10s, 6.47s, 4.03s, 5.85s, 3.39s, 2.39s, 2.16s, 2.82s

        Show
        Robert Muir added a comment - unpatched: 62.87s, 14.08s, 6.05s, 41.36s, 3.90s, 8.14s, 12.50s, 16.09s, 7.02s, 7.18s, 125.19s, 37.17s patched: 3.03s, 3.76s, 3.22s, 4.20s, 2.10s, 6.47s, 4.03s, 5.85s, 3.39s, 2.39s, 2.16s, 2.82s
        Hide
        Chris Male added a comment -

        Oh right I understand what you mean by currently now.

        Show
        Chris Male added a comment - Oh right I understand what you mean by currently now.
        Hide
        Robert Muir added a comment -

        This patch fixes the variance (as mentioned above)

        Show
        Robert Muir added a comment - This patch fixes the variance (as mentioned above)
        Hide
        Chris Male added a comment -

        I havent looked at the test much but I imagine it probably randomly generates groups, so maybe the number of groups heavily influences the time.

        Show
        Chris Male added a comment - I havent looked at the test much but I imagine it probably randomly generates groups, so maybe the number of groups heavily influences the time.
        Hide
        Robert Muir added a comment -

        This patch seems to stabilize the runs (currently its sometimes like 5 seconds, other times like 2 minutes!)

        Show
        Robert Muir added a comment - This patch seems to stabilize the runs (currently its sometimes like 5 seconds, other times like 2 minutes!)
        Hide
        Robert Muir added a comment -

        Martijn, I don't think the current number is ok for nightly builds either.

        This test took more than an hour last night!

        Show
        Robert Muir added a comment - Martijn, I don't think the current number is ok for nightly builds either. This test took more than an hour last night!
        Hide
        Martijn van Groningen added a comment -

        I think lowering the number of search and index iterations would be fine for regular builds and nightly builds can keep the current number of maximum iterations.

        Show
        Martijn van Groningen added a comment - I think lowering the number of search and index iterations would be fine for regular builds and nightly builds can keep the current number of maximum iterations.

          People

          • Assignee:
            Robert Muir
            Reporter:
            Robert Muir
          • Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved:

              Development