1 Reply Latest reply on Jan 23, 2018 5:26 PM by pferraro

    Kubernetes Cluster discovery support on Openshift for WildFly 11

    vikrant02

      Hi,

       

      I am deploying keycloak cluster which is running WildFly 11 on OpenShift environment.

       

      Currently WildFly 11 doesn't provide out-of-box support for native kubernetes api for cluster discovery and we are using FILE_PING instead for discovery which is not as efficient and reliable as KUBE_PING discovery. Is there a way to use KUBE_PING 1.0.5 with wildfly 11 or any other efficient alternative of FILE_PING on OpenShift.

       

      Thanks,

      Vikrant

        • 1. Re: Kubernetes Cluster discovery support on Openshift for WildFly 11
          pferraro

          WF11 uses JGroups 3.6.x, so you'd need to use a compatible release of KUBE_PING (e.g. 1.0.x appears to require JGroups 4.0.x).  Assuming you find one, package it as a module and reference it in your jgroups stack using its module name.

          e.g.

          <protocol type="kubernetes.KUBE_PING" module="org.jgroups.kubernetes"/>

          or

          /subsystem=jgroups/stack=foo/protocol=kubernetes.KUBE_PING:add(module=org.jgroups.kubernetes", add-index=N){allow-resource-service-restart=true}

           

          WF12 will use JGroups 4.0.x, and will likely include a module containing KUBE_PING by default, so this will be easier very soon.