All databases missing after MySQL cluster restart
Posted by:
webmaster
Date: September 18, 2006 03:39AM
Hello,
I have installed MySQL 5.1.11 beta on 4 AMD64 Debian boxes. All four nodes are running mysqld, two of the nodes are running ndbd, and the two remaining nodes are management servers (with mysqld). I am using statically-defined nodeid's in mysql.ini All connectstring's also use statically defined nodeid's. I am using 2 replicas.
I got the cluster up and running successfully and created several databases and tables. I was able to access and modify the database from each of the four nodes. Changes made in one node were reflected in other nodes.
However, when I shutdown the cluster, rebooted the nodes, and started MySQL cluster, all of the databases that I had created disappeared! This happens every time I create a cluster, add databases, and restart the cluster. If I restart just a single node, then all of the databases are missing just from that one node.
Grep reports that the missing databases exist in SO.TableList, So.FragLog, and T[1,2,5].Data
ndb_waiter reports:
<snip>
Connecting to mgmsrv at nodeid=3,host=node00014,host=node00015
State node 3 STARTED
State node 4 STARTED
Waiting for cluster enter state STARTED
NDBT_ProgramExit: 0 - OK
</snip>
ndb_mgm -e show reports that all mysqld, ndb_mgmd, and ndbd nodes are up and connected.
'SHOW ENGINES' reports that the ndbcluster is the default mysql engine and that it is enabled.
The ndb logfiles are not reporting any errors.
I tried doing 'FLUSH TABLES', however that did not help.
This problem is consistently reproducible whenever I reinitialize the cluster, create new databases and tables, and restart the cluster.
Why are all of my ndb databases disappearing from mysqld whenever I restart the cluster (or just a single node)?
Thanks in advance