[ https://issues.redhat.com/browse/WFLY-9641?page=com.atlassian.jira.plugin... ] Gregory Hall edited comment on WFLY-9641 at 12/27/19 11:31 AM: --------------------------------------------------------------- This also affects Wildfly 16 (I'm trying to cluster keycloak 6.0.1 which is based on Wildfly 16) I'm running in a docker container hosted in AWS cloud where multicast isn't an option. I've configured external addresses on all the JGroups TCPPING protocol options. modcluster has been removed since we're using nginx load balancer in ip_hash mode. There are other timeout issues when nginx fails over to a different node despite the cluster view updating. Not sure what the best tactic as using 0.0.0.0 inside a container is definitely preferred and attempts to bind either the 172 or 10 ifc are failing to allow external access. The basic pretense that the ejb cluster would second guess routability through an inadequate test sounds wrong. Due to NAT with/without docker containers and networks, the best practice would be to try to connect and if you throw a NoRouteToHost, deal with that. was (Author: gregoryhall994): This also affects Wildfly 16 (I'm trying to cluster keycloak 6.0.1 which is based on Wildfly 16) I'm running in a docker container hosted in AWS cloud where multicast isn't an option. I've configured external addresses on all the JGroups TCPPING protocol options. Not sure what the best tactic as using 0.0.0.0 inside a container is definitely preferred. The basic pretense that the ejb cluster would second guess routability through an inadequate test sounds wrong. Due to NAT with/without docker containers and networks, the best practice would be to try to connect and if you throw a NoRouteToHost, deal with that. -- This message was sent by Atlassian Jira (v7.13.8#713008)