one of the nodes failed then the whole cluster went down.

shutdown mysql on all nodes

cat /var/lib/mysql/grastate.dat

check which node has the highest seq number

boot strap that node

service mysql bootstrap-pxc

then start the next node but before that

mv /var/lib/mysql/grastate.dat ~

service mysql start

this will start a full resync

wait till it finishes then start the same process on the next node.

 

https://www.percona.com/doc/percona-xtradb-cluster/5.5/manual/bootstrap.html