Inconsistent data in cluster of 3 nodes

I am started to use in production the Percona XtraDB cluster. I checked the number of rows in each table in each node and the total is different in each node. All node have:

wsrep_ready ON

and wsrep_local_state_uuid is equal to wsrep_cluster_state_uuid

It is clear that the cluster has made data inconsistent and it does not keep things on sync. How should I processed to fix the issues? How is it possible to have a different number of rows in the nodes and the cluster replication seems ok?

Actually I found that this query


returns sort of casual numbers, so if I run it 10 times, most of the times the number of rows are different. I saw that two tables have an issue: when a new row is inserted, the change is not replicated to the other two hosts. What would be the cause? How to fix it?

The DB should had only InnoDB tables, but the two tables where MyIsam!