Uploaded image for project: 'JGroups'
  1. JGroups
  2. JGRP-1794

OOBTest method testNonBlockingUnicastOOBMessage fails to correctly receive all OOB mssages

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Major Major
    • 3.2.13
    • 3.2.13
    • None

      This test fails regularly on multiple platforms. It doesn't fail all the time, but perhaps on 5/10 platform executions.

      The test consists of a sender and a receiver. The receiver B will receive OOB messages but blocks on a latch for 25 seconds when it receives a non-OOB message. After 25 seconds, it unblocks and continues receiving

      The test does this:

      • send 1 regular message from A -> B
      • send 9 OOB meesages from A -> B
      • wait 20 seconds for messages to arrive
      • check that all 9 OOB messages have arrived
      • unblock the latch
      • check that all 10 message have arrived

      The test failures all involve the first 9 OOB messages not arriving completely. For example, sample lists of received messages before the first 20 second interval are:

      [3,2,6,8,9,10,4]
      [2,3,4,8,9,5]
      [2,3,4,8,7,9]
      [2,3,5,4,6]

      So, it appears that either the messages are slow to arrive or not arriving at all.

      A correct result looks like this:

      list = [2, 4, 3, 7, 8, 9, 10, 5, 6]
      [main]: releasing latch
      list = [2, 4, 3, 7, 8, 9, 10, 5, 6, 1]
      

              rhn-engineering-bban Bela Ban
              rachmato@redhat.com Richard Achmatowicz
              Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

                Created:
                Updated:
                Resolved: