Uploaded image for project: 'Ignite'
  1. Ignite
  2. IGNITE-10354

Failing client node due to not receiving metrics updates

Attach filesAttach ScreenshotVotersStop watchingWatchersCreate sub-taskLinkCloneUpdate Comment AuthorReplace String in CommentUpdate Comment VisibilityDelete Comments
    XMLWordPrintableJSON

Details

    • Bug
    • Status: Resolved
    • Major
    • Resolution: Fixed
    • 2.6
    • 2.8
    • clients
    • None

    Description

      In some cases after the coordinator change, the client node can be failed before it can establish a connection to another server from the cluster.

      [2018-11-21 12:21:45,769][WARN ][tcp-disco-msg-worker-#15%server-b%][TestTcpDiscoverySpi] Failing client node due to not receiving metrics updates from client node within 'IgniteConfiguration.clientFailureDetectionTimeout' (consider increasing configuration property) [timeout=10000, node=TcpDiscoveryNode [id=dc739711-f685-45e8-9017-1f91b1d86c8c, addrs=[0:0:0:0:0:0:0:1, 10.0.75.1, 127.0.0.1, 192.168.1.51, 192.168.192.1], sockAddrs=[/0:0:0:0:0:0:0:1:0, LAPTOP-6FN8RAOS/10.0.75.1:0, /127.0.0.1:0, /192.168.192.1:0, /192.168.1.51:0], discPort=0, order=2, intOrder=2, lastExchangeTime=1542774105666, loc=false, ver=2.4.0#20180830-sha1:345c0a7c, isClient=true]]
      [2018-11-21 12:21:45,791][INFO ][tcp-client-disco-msg-worker-#10%client%][TestTcpDiscoverySpi] Client node disconnected from cluster, will try to reconnect with new id [newId=46812956-2fc4-4b74-9909-d523a547ba0e, prevId=dc739711-f685-45e8-9017-1f91b1d86c8c, locNode=TcpDiscoveryNode [id=dc739711-f685-45e8-9017-1f91b1d86c8c, addrs=[0:0:0:0:0:0:0:1, 10.0.75.1, 127.0.0.1, 192.168.1.51, 192.168.192.1], sockAddrs=[/0:0:0:0:0:0:0:1:0, LAPTOP-6FN8RAOS/10.0.75.1:0, /127.0.0.1:0, /192.168.192.1:0, /192.168.1.51:0], discPort=0, order=2, intOrder=0, lastExchangeTime=1542774104031, loc=true, ver=2.4.0#20180830-sha1:345c0a7c, isClient=true]]
      

      It looks like a race condition.

      Steps to reproduce:

      1. Start server A.
      2. Start client.
      3. Start server B.
      4. Stop server A.

      If add Thread.sleep(10000) between (3) and (4) then the client node won't be disconnected from the cluster.

      Reproducer is attached ClientDisconnectedTest.java.

      Attachments

        Activity

          This comment will be Viewable by All Users Viewable by All Users
          Cancel

          People

            guseinov Roman Guseinov
            guseinov Roman Guseinov
            Votes:
            0 Vote for this issue
            Watchers:
            4 Stop watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Slack

                Issue deployment