Re: NDB Cluster circular replication

[Mikael Ronström]

Best guess I have is that you have started all three slaves in
the cluster. This means that there are 3 MySQL servers trying to
replicate the same changes into the cluster, they will collide and
have a high likelihood of reaching deadlocks and other issues.

In a dual replication setup you need to setup two slaves, but only one
of them can be started. If the started slave fails, then you need to
start the second slave, but before doing so you need to discover from
where in the binlog it needs to start. This is described in the docs
I think and in books about MySQL Cluster.