MySQL Bugs: #19864: Cluster restarted, but data is inconsistent across data nodes

Bug #19864	Cluster restarted, but data is inconsistent across data nodes
Submitted:	17 May 2006 2:26	Modified:	17 May 2006 3:43
Reporter:	David Rusenko	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Server	Severity:	S1 (Critical)
Version:	5.0.18	OS:	Linux (SuSE Linux 9.1)
Assigned to:		CPU Architecture:	Any

Description:
One of our systems hung, so we had to perform a reboot. After the reboot, the ndbd node was reconnected, but none of the data would show up on this half of the cluster, even though it was shown as "connected" in the ndb_mgm console. The node was restarted with "--initial" as well, to no avail. At this point, there is a disconnect across nodes, where one node has the correct data, and the other, strangely enough, has one table's worth of metadata, and nothing else. "FLUSH TABLES" does not remedy the situation on the "broken" node. 

How to repeat:
kill one ndbd process, and restart.

Suggested fix:
None known or available.

hile reading through other posts on the MySQL forums, I noticed one user who had the NDBCLUSTER engine set as DISABLED. I checked this on our node, and sure enough, it was set to disabled as well. It seems that, although there were free API nodes available, the ndb_mgmd server was somehow not making these available. A quick restart of the ndb_mgmd server and API node in question seems to have solved the issue.