jgroups errors in clustered setup

We have a number of xwiki servers set up in a clustered environment. Occasionally, we are not able to "bounce" our servers - we restart the servers one by one, and get many of the following error: ERRORS>> ERROR [org.jgroups.protocols.pbcast.ClientGmsImpl] <08:30:24,559> suspect() should not be invoked on an instance of org.jgroups.protocols.pbcast.ClientGmsImpl <UpHandler (GMS)> We upgraded our jgroups to 2.2.9, and this has significantly reduced how often this happens, but it still does happen.When this happens, we need to bring down all the servers and restart. Has anyone encountered this issue ? What is the recommended version of jgroups and configuration ? Is there a recommendation for how to restart xwiki servers while achieving 24x7 uptime witgh a cluster ? Thank you in advance Andy