Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-6947591

JDK ParallelGCThreads does not honor cpu/zone capping issue

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Cannot Reproduce
    • Icon: P4 P4
    • None
    • 6u19
    • hotspot
    • gc
    • sparc
    • solaris_10

      The systema is an M9000 / SPARC
      with 64 total cpu's


      1. In the example, the zone had a CPU cap of 40, but it appears
           the CPU cap was ignored and the Ergonomics routine went off the total
           number of physcial CPUs and ignored the Cap. (see below)
           This appears wrong and should be fixed.

      Following zones are currently running on wdcsn230a:
              Total processors available to all zones:64

      Zone Name CPU Capping Resource Pool Pool size Swap Mem
      ---------------------------------------------------------------------------------------
      wdcsn251a - pool_default 64 - -
      wdcsn241a - pool_default 64 - -
      wdcsn272a - pool_default 64 - -
      wdcsn231a - pool_default 64 - -
      wdcsn277a - pool_default 64 - -
      wdcsn232a 40.00 pool_default 64 - -
      wdcsn189a - pool_default 64 - -

      sr1-ubur-21:/net/cores/cores/dir31/72706004 348 % grep -ic GCTaskThread pstack_before.txt-dem
      43

            jmasa Jon Masamitsu (Inactive)
            dvanero Dennis Van Eron (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Created:
              Updated:
              Resolved:
              Imported:
              Indexed: