'Got error 0 'No error' from NDBCLUSTER (no replies)

I am attempting to migrate some tables from MyISAM to NDB but I'm running in to an error on the replication thread (normal query-based replication, not the cluster replication):

121025 21:16:36 [ERROR] Slave SQL: Error 'Got error 0 'No error' from NDBCLUSTER' on query. Default database: 'accounts'. Query: 'UPDATE heartbeat SET ts = NOW() WHERE id = 1', Error_code: 1296
121025 21:16:36 [Warning] Slave: Got error 0 'No error' from NDB Error_code: 1296
121025 21:16:36 [Warning] Slave: Got error 0 'No error' from NDBCLUSTER Error_code: 1296

This occurs after I run:

ALTER TABLE heartbeat TABLESPACE ts_1 STORAGE DISK ENGINE=NDBCLUSTER;

and wait a second for replication to relay that UPDATE statement.

So far I have tried the following:

* Restarting all data+management+API nodes
* Re-creating the NDBCLUSTER table from scratch
* Resetting all of the data nodes one at a time (ndbmtd --initial)
* Upgrading from 7.2.6 to 7.2.8 (and then recreating the table again because of the hashmap change referred to here: http://forums.mysql.com/read.php?25,566540,566630#msg-566630)

I can only get replication running again if I convert the table back to MyISAM. Also, I can run the query manually. It's just failing when it runs on the replication thread.

The only thing I haven't tried is kill -9+rm -rf on every cluster node and starting over from zero.

Has anyone else run in to this problem, and if so, what was your solution? What should I look at? I am running 5.5.27 w/ ndb 7.2.8