0 Replies Latest reply on Oct 18, 2017 6:22 PM by Charles Morris

    JDBC_PING Failure detection and no physical address error

    Charles Morris Newbie

      Hi,

       

      I'm not sure if this is the correct place to post this or not...

       

      Using Tomcat 8.5.16, with a proprietary application that comes bundled with jgroups-3.6.11.Final.jar in the WEB-INF/lib folder.

       

      We have a 2 node cluster setup using JDBC_PING, which appears to be working:

       

      -------------------------------------------------------------------

      GMS: address={{hostname-1}}-15359, cluster=ClusterGroup-4viEDGGQuxg8rorYM5WY8XgYFlQ=, physical address=10.3.82.79:7800

      -------------------------------------------------------------------

      |

      | Cluster Started 

      |

      | Host identified as: {{hostname-1}}-15359

      | Cluster identified as: {{hostname-2}}-36886

      | Additional Cluster Info: [{{hostname}}-36886|1] (2) [{{hostname-1}}-36886, {{hostname-2}}-15359]

      |

      YF:2017-10-19 09:01:49: INFO (SystemTaskManager:setEnabled) - System Task Manager has been disabled

      |

      | Session Replication: false

      |

      | Cluster Management classes have been initialised.

      |

      ========================================================================

       

      That's great, and I can see the two entries in the database.

       

      For failure detection we have the following settings:

       

      <FD_SOCK

        start_port="7979"

      />

       

      So we can poke a hole in the security group (AWS), and they can 'ping' each other. I'm not really sure what that is for though; the clustering seems to work fine without that firewall rule: If a node dies, the other one takes over, and so on. So what situation would that be useful in?

       

      Anyway, here is the error we get on startup of the secondary node:

       

      WARNING [TransferQueueBundler,ClusterGroup-4viEDGGQuxg8rorYM5WY8XgYFlQ=,i001-yf008-prod003-36886] org.jgroups.protocols.TP.sendToSingleMember JGRP000032: {{hostname-1}}-36886: no physical address for a3aeb8de-517a-299c-4adc-c404ea2cba5e, dropping message

       

      Is that just a one time warning, and then it finds the physical address? I'm not entirely sure if it is working or not. This message is AFTER the cluster initialisation.

       

      Thanks for your help.