MySQL Bugs: #51121: Segmentation fault in data node

Bug #51121	Segmentation fault in data node
Submitted:	11 Feb 2010 21:15	Modified:	17 Mar 2010 15:17
Reporter:	Candice Savino	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S1 (Critical)
Version:	mysql-5.1-telco-7.0	OS:	Linux
Assigned to:	Jonas Oreland	CPU Architecture:	Any
Tags:	5.1.39-ndb-7.0.9-cluster-gpl

Description:
This happens randomly during normal use.  The cluster goes down and the ndbd node will not come back up without using the initial flag and starting from scratch.

I have attached ndb_error_reporter document.  

How to repeat:
There were auth requests/writes to this db and retrieval/update of information.

Hi,

What seems to be consistent is that it crashes when doing insert into table with id 39.

Could you run "ndb_show_tables", find out which table that is, and
upload create table statement for that table.

If you we're worried about privacy concerns you can
1) upload it privatly
2) obfuscate table/column names (i'm only interested in structure)

/Jonas

Below is the table structure for table 39.  I changed the table names and columns.

CREATE TABLE `TABLE_X` (
  `AUTO_UUID` mediumint(9) NOT NULL AUTO_INCREMENT,
  `UUID` varchar(16) CHARACTER SET latin1 NOT NULL,
  `XX_ID` int(11) NOT NULL,
  `XXX_ID` int(11) NOT NULL,
  `BIT` tinyint(1) DEFAULT NULL,
  PRIMARY KEY (`AUTO_UUID`),
  UNIQUE KEY `UNIQUE_KEY` (`UUID`,`XX_ID`,`XXX_ID`)
) ENGINE=ndbcluster AUTO_INCREMENT=2418 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;

Thanks!

Hi,

the table definition didn't give much extra info.
sorry :-(

maybe you can enable general log on mysqld
to see if the query that causes the crash is possible to identify.

/Jonas

You're correct, table 111 is shown as last insert.
I'm not aware of any bugs corresponding to such a table definition.
Could you do a "all report memory" too see if there can be some error handling
case wo/ out of memory.

And also "ndb_desc -p -d <db> TABLE" (which should show #rows etc)

We rebuild our environment and this seg fault issue has gone away so far.  Looks like this was an environmental issue.

Thanks for your help.