Comment 3 for bug 1037380

The mailing list link does not seem to be related.

The crash is due to data inconsistency in the cluster, Galera has recognized that this node does not have same data content as other nodes in the cluster and does emergency shutdown.

Note, that when the table was of type MyISAM, it was not subject to replication (well, unless you have deliberately configured wsrep_replicate_myisam). So whatever you write to this table will remain local in respective node. Then, when you alter the table to InnoDB, it will become a replicated table and if nodes contained different data at this point, it will lead to sure crash.

Can you confirm that table contents were consistent before the alter to innodb was issued?