Issues with a new Cluster - ERROR 1114 (HY000): The table 'xyz' is full
Posted by: Brad R
Date: December 03, 2013 04:35PM
First off, thanks for reading, second, I have a fledgling MySQL Cluster that I setup, it consists of two servers running according to the mysql website (I have tested creating new test tables and they work FLAWLESSLY), up until a few hours ago I was chugging along with no issues and quite happy.
After doing all of my initial tests, confirming that a reboot comes up cleanly and services run without any interactions from me, I took one of my bigger clients databases, and started converting engine portions over to NDBCLUSTER, when I was hit with a rather large headache that I have been racking over google trying to figure out..
Here is the query I am throwing to mysql to process:
CREATE TABLE `ibf_admin_logs` (
`id` bigint(20) NOT NULL AUTO_INCREMENT,
`member_id` int(10) DEFAULT NULL,
`ctime` int(10) DEFAULT NULL,
`ip_address` varchar(46) DEFAULT NULL,
`appcomponent` varchar(255) NOT NULL DEFAULT '',
`module` varchar(255) NOT NULL DEFAULT '',
`section` varchar(255) NOT NULL DEFAULT '',
`do` varchar(255) NOT NULL DEFAULT '',
PRIMARY KEY (`id`),
KEY `ctime` (`ctime`),
KEY `ip_address` (`ip_address`)
) ENGINE=NDBCLUSTER DEFAULT CHARSET=latin1 AUTO_INCREMENT=3069;
It being an existing table, I just changed out the engine and submitted it to the cluster, however I am getting this error:
ERROR 1114 (HY000): The table 'ibf_admin_logs' is full
From what instances I found, other users have been able to fix this by increasing their DataMemory directive to something larger, however, since this system has essentially KBs of data on it, I still proceeded to tune it up to what we were looking to have it at (the server has 32GB presently, so I tuned it to 20480MB of RAM in the config file, the configuration portion can be found here, however, I still get the error. If I change it to MyISAM, which was what it was prior, it creates it fine, but obviously no replication to the other cluster members...
NoOfReplicas = 2 #Number of replicas, default:2
DataMemory = 20480MB
IndexMemory = 512M
BackupMemory = 512M
DataDir = /var/lib/mysql-cluster #directory for storing management data
MaxNoOfTables = 1024
I am honestly out of ideas, this is my last big hurdle before I can go to my engineering core and say we have a functioning cluster, let's get off of the individual boxes in favor of this!~
One thing to note, I DID import a cacti database that was at 5.8MB, i was able to import ALL of the tables except for 2, they were at the end and gave me errors until I removed the MyISAM tables and readded as NDBCLUSTERed tables.. Is there a setting somewhere that correlates with a 6MB size for this?
Anyone's expertice would be appreciated.
Edited 1 time(s). Last edit at 12/03/2013 04:54PM by Brad R.