[ https://issues.redhat.com/browse/JGRP-2486?page=com.atlassian.jira.plugin... ] Bela Ban edited comment on JGRP-2486 at 6/26/20 8:08 AM: --------------------------------------------------------- This wouldn't happen if you used {{UDP}} or {{TCP_NIO2}}. Also, if you disabled the NIC, TCP itself would block after a while, so what's the use case here? Note that you could use {{RED}} to drop messages if the queue starts getting full... The surviving node should suspect and exclude the dead node. Note that I don't recommend FD, I suggest use FD_ALL2 or FD_ALL3 instead; there have been some nice improvements in the latter, e.g. skipping the sending of heartbeats if real traffoc is received etc was (Author: belaban): This wouldn't happen if you used {{UDP}} or {{TCP_NIO2}}. Also, if you disabled the NIC, TCP itself would block after a while, so what's the use case here? Note that you could use {{RED}} to drop messages if the queue starts getting full... -- This message was sent by Atlassian Jira (v7.13.8#713008)