Zombie Node

I have a 6 node cluster that was running well ,
Unfortunately the other day one of the servers had a lockup , still pingable but not reachable ,
Until I killed the node the entire cluster was locked and was unable to make any queries ,
Version I am using is Percona-XtraDB-Cluster 56-5.6.22-25.8.978
once the Node was disabled the services retuned to normal

Has anybody experienced this before and a possible solution to stop it happening again .



By ‘unable to make any queries’ do you mean SELECTs too?
If one node is locked in a cluster, which means it’s still a valid member, has communication to other nodes working, but for some reason it is unable to apply writes, a Flow Control mechanism will fire up, which eventually will pause writes on all other nodes. This is in order to prevent too much replication lag in a cluster. See more details here: