Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-8261614

JDK-8252522 causes big execution time increase for open/test/hotspot/jtreg/vmTestbase/nsk/monitoring

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Not an Issue
    • Icon: P4 P4
    • 18
    • 16
    • core-svc

      I run my personal stress kit on promoted JDK bits and I
      noticed a big execution time increase of the VM/NSK M&M
      tests starting in jdk-16+15.

      Here's the end of my bisection analysis for the changesets
      that went into jdk-16+15:

      redo_7:
      @closed changeset: 4b8a3d (HEAD, tag: jdk-16+15) 8252814: Disable client tests until lab systems back on line
      @open changeset: 1262ae36af0 (HEAD) 8252679: Two windows specific FileDIalog tests may fail on some Windows_Server_2016_Standard

      $ elapsed_times jdk-16+14_redo_7/logs.linux-x86_64/*
      jdk-16+14_redo_7/logs.linux-x86_64/start_run.log 0 seconds
      jdk-16+14_redo_7/logs.linux-x86_64/do_java_bld.log 29 minutes 26 seconds
      jdk-16+14_redo_7/logs.linux-x86_64/do_monitoring_tests-release.log 31 minutes 16 seconds
      jdk-16+14_redo_7/logs.linux-x86_64/do_monitoring_tests-fastdebug.log 37 minutes 35 seconds
      jdk-16+14_redo_7/logs.linux-x86_64/do_monitoring_tests-slowdebug.log 1 hours 1 minutes 5 seconds

      And here's the results for the problematic changeset:

      redo_8:
      @closed changeset: 4b8a3d (HEAD, tag: jdk-16+15) 8252814: Disable client tests until lab systems back on line
      @open changeset: 5f76deb2e06 (HEAD) 8252522: nsk/share/test/StressOptions should multiple stressTime by jtreg's timeout-factor

      $ elapsed_times jdk-16+14_redo_8/logs.linux-x86_64/*
      jdk-16+14_redo_8/logs.linux-x86_64/start_run.log 0 seconds
      jdk-16+14_redo_8/logs.linux-x86_64/do_java_bld.log 29 minutes 16 seconds
      jdk-16+14_redo_8/logs.linux-x86_64/do_monitoring_tests-release.log 1 hours 42 minutes 4 seconds
      jdk-16+14_redo_8/logs.linux-x86_64/do_monitoring_tests-fastdebug.log 2 hours 42 minutes 0 seconds
      jdk-16+14_redo_8/logs.linux-x86_64/do_monitoring_tests-slowdebug.log 5 hours 54 minutes 38 seconds

      release bits went from 31 minutes → 102 minutes
      fastdebug bits went from 38 minutes → 162 minutes
      slowdebug bits went from 61 minutes → 355 minutes

      So for release bits, we have:

      (102 - 31) / 31 * 100 => 229% increase

      for fastdebug bits we have:

      (162 - 38) / 38 * 100 => 326% increase

      and for slowdebug bits we have:

      (355 - 61) / 61 * 100 => 481% increase

            Unassigned Unassigned
            dcubed Daniel Daugherty
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

              Created:
              Updated:
              Resolved: