2 Replies Latest reply on Jun 15, 2016 11:06 AM by pferraro

    Wildfly 10 HA Clustering on AWS does not work

    xxlali
    xxlali Jun 14, 2016 2:19 AM

      I am trying to set up an HA Clustering on Amazon Web Service with Wildfly-10. My standalone-ha.xml configuration is

       

       

          ...

          <subsystem xmlns="urn:jboss:domain:jgroups:4.0">

          <channels default="ee">

              <channel name="ee" stack="s3ping"/>

          <channels>

          <stacks>

                 ...         

              <stack name="s3ping">

                 <transport type="TCP" socket-binding="jgroups-tcp" diagnostics-socket-binding="jgroups-diagnostics"/>

                     <protocol type="S3_PING">

                         <property name="access_key">

                             <%= @s3_access_key %>

                         </property>

                         <property name="secret_access_key">

                             <%= @s3_secret_access_key %>

                         </property>

                         <property name="prefix">

                             <%= @s3_bucket %>

                         </property>

                         <property name="timeout">

                             60000

                         </property>

                     </protocol>

                     <protocol type="MERGE2"/>

                     <protocol type="FD_SOCK" socket-binding="jgroups-tcp-fd"/>           

                     <protocol type="FD"/>

                     <protocol type="VERIFY_SUSPECT"/>

                     <protocol type="BARRIER"/>

                     <protocol type="pbcast.NAKACK"/>

                     <protocol type="UNICAST2"/>

                     <protocol type="pbcast.STABLE"/>

                     <protocol type="pbcast.GMS"/>

                     <protocol type="UFC"/>

                     <protocol type="MFC"/>

                     <protocol type="FRAG2"/>

                 </stack>

          </stacks>

          </subsystem>

          ...

       

       

      With very similar configuration I am able to cluster with Wildfly 8 and Wildfly 8.2. But with Wildfly 10, I am not able to succeed it.

       

       

      Also I have added `<distributable/>` to my `web.xml.


      I have .two machines. When one of them down, the other one says;

       

            INFO  [org.infinispan.remoting.transport.jgroups.JGroupsTransport] (Incoming-4,ee,node1) ISPN000094: Received new cluster view for channel server: [node1|2] (1) [node1]

            INFO  [org.infinispan.remoting.transport.jgroups.JGroupsTransport] (Incoming-4,ee,node1) ISPN000094: Received new cluster view for channel web: [node1|2] (1) [node1]

            INFO  [org.infinispan.remoting.transport.jgroups.JGroupsTransport] (Incoming-4,ee,node1) ISPN000094: Received new cluster view for channel ejb: [node1|2] (1) [node1]

            INFO  [org.infinispan.remoting.transport.jgroups.JGroupsTransport] (Incoming-4,ee,node1) ISPN000094: Received new cluster view for channel hibernate: [node1|2] (1) [node1]


      As I understand here, it gets session info from the down machine. But, when client makes a new request its session id changes and the system redirects user the login page.

       

      How can I make the client keep its session id and continue to use system.

       

       

      Thanks.