Bug #7799 Cluster crash : failed ndbrequire, fault id: 2341
Submitted: 11 Jan 2005 12:34 Modified: 11 Jan 2005 12:39
Reporter: Simon Detheridge Email Updates:
Status: Duplicate Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S3 (Non-critical)
Version:4.1.8 OS:Linux (Linux (gentoo))
Assigned to: Assigned Account CPU Architecture:Any

[11 Jan 2005 12:34] Simon Detheridge
Description:
Running ndbcluster with two database nodes and one mgm node. NoOfReplicas=2

The cluster crashes when performing a SELECT, with the following errors on the database nodes.

The nodes die with the following errors:

> Date/Time: Monday 10 January 2005 - 14:25:05
> Type of error: error
> Message: Internal program error (failed ndbrequire)
> Fault ID: 2341
> Problem data: DbtcMain.cpp
> Object of reference: DBTC (Line: 1321) 0x0000000a
> ProgramName: /usr/sbin/ndbd
> ProcessID: 15037
> TraceFile: /var/lib/mysql-cluster/ndb_12_trace.log.4
> ***EOM***

> Date/Time: Monday 10 January 2005 - 14:26:46
> Type of error: error
> Message: Internal program error (failed ndbrequire)
> Fault ID: 2341
> Problem data: DbtcMain.cpp
> Object of reference: DBTC (Line: 1321) 0x0000000e
> ProgramName: ndbd
> ProcessID: 3525
> TraceFile: /var/lib/mysql-cluster/ndb_13_trace.log.1
> ***EOM***

I have posted the gzipped tracefiles at http://www.symdata.com/simon/

How to repeat:
The server seems to crash when doing a join accross two tables, created as follows:

CREATE TABLE `t1` (
  `mid` bigint(11) NOT NULL auto_increment,
  `uid` int(11) NOT NULL default '0',
  `aid` bigint(11) NOT NULL default '0',
  `subj` varchar(255) NOT NULL default '',
  `tdate` datetime NOT NULL default '0000-00-00 00:00:00',
  `text` mediumtext NOT NULL,
  `native` mediumtext NOT NULL,
  `mod mediumtext NOT NULL,
  `subjmod` text,
  PRIMARY KEY  (`mid`),
  KEY `maildate_key` (`uid`,`tdate`),
  KEY `subject_key` (`uid`,`subj`),
  KEY `uid_key` (`uid`)
) ENGINE=ndbcluser DEFAULT CHARSET-latin1

and

CREATE TABLE `t2` (
  `aid` bigint(20) NOT NULL auto_increment,
  `name` varchar(255) NOT NULL default '',
  `add` varchar(255) NOT NULL default '',
  `uid` int(11) NOT NULL default '0',
  `image` mediumblob,
  PRIMARY KEY  (`aid`),
  KEY `uid` (`uid`),
  KEY `uid_2` (`uid`,`name`),
  KEY `uid_3` (`uid`,`address`)
) ENGINE=ndbcluster DEFAULT CHARSET=latin1

The query looks something along the lines of:

SELECT t2.name AS name , 
	t1.tdate AS tdate , 
	t1.subj AS subj , 
	t1.mid AS mid 
	FROM t1 JOIN t2 
	ON t1.aid = t2.aid 
	WHERE t1.uid=$uid ORDER BY t1.tdate DESC
[11 Jan 2005 12:39] Jonas Oreland
I reported the bug myself as 7798,
The fix will be in 4.1.9