Storage node partitioning and arbitration
In this recipe, we explore what happens when a MySQL Cluster has its storage nodes split into two groups that cannot communicate, but each of which has a full set of cluster data. We will look at this with a practical example, with the explanation of the process in the There's more... section.
In our example lab, nodes
2 make up nodegroup
0 and nodes
4 make up nodegroup
1 (look at the output in the recipe Single storage node failure from the
SHOW command inside
ndb_mgm to see this). In earlier recipes, we have covered what happens if we shut down any combination of nodes, but not what occurs if one node in each nodegroup is isolated from the rest of the cluster.