[ https://issues.jboss.org/browse/JGRP-2253?page=com.atlassian.jira.plugin.... ] Bela Ban commented on JGRP-2253: -------------------------------- I can't really see something other than the SUSPECT/UNSUSPECT cycles in your logs, which are suspicious and point to FD_SOCK not being able to connect to the destination it is supposed to monitor: {noformat} 21:52:16.021 TRACE 1108 — [jgroups-13,cluster,ip-143-60500] FD_SOCK : - - ip-143-60500: received SUSPECT message from ip-48-43121: suspects=[ip-143-60500] 21:52:16.022 TRACE 1108 — [jgroups-13,cluster,ip-143-60500] FD_SOCK : - - ip-143-60500: received SUSPECT message from ip-48-43121: suspects=[ip-163-65450] 21:52:16.024 TRACE 1108 — [jgroups-3,cluster,ip-143-60500] FD_SOCK : - - ip-143-60500: received UNSUSPECT message from ip-48-43121: mbrs=[ip-143-60500] 21:52:16.025 TRACE 1108 — [jgroups-4,cluster,ip-143-60500] FD_SOCK : - - ip-143-60500: received UNSUSPECT message from ip-48-43121: mbrs=[ip-163-65450] {noformat} My bet is that this is caused by incorrect ports, which causes FD_SOCK to malfunction. I'm also not sure that termination on EC2/AWS closes the sockets of a process, or whether termination is more like a power-down / pull-the power-plug like behavior. I seem to recall that this was the case. If so, FD_SOCK would not kick in, but FD would instead... -- This message was sent by Atlassian JIRA (v7.5.0#75005)