MySQL Bugs: #25680: server crashes

Bug #25680	server crashes
Submitted:	17 Jan 2007 14:57	Modified:	12 May 2007 17:01
Reporter:	aligi piccin	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Server	Severity:	S1 (Critical)
Version:	5.0; 5.1	OS:	Windows (W2k3 64bit)
Assigned to:		CPU Architecture:	Any
Tags:	crash, INSERT IGNORE, server, win64

Description:
My program (written in DELHPI, using ODBC to connect to MySQL Server) runs consecutively some queries like this, on different tables:

INSERT IGNORE INTO mytable_1_old SELECT * FROM mytable_1 WHERE myfield ="x"

All test done on the same machine: a W2k3 server 64bit
No problem when using the 32bit MySQLserver but a lot of server crashes when using 64bit MySQLserver, so i had to abandon the 64bit MySQLserver and use the 32bit MySQLserver.
In both cases i used the same databases, with the same data tables, same configuration, same everything. 

How to repeat:
Mytable_n and mytable_n_old have the same structure
Mytable_n are usually "big" (>1,000,000 rows) 
The number of rows affected by the queries can be "big" (>100.000)

Thank you for a problem report. Please, send the exact versions, 5.x.y, that crash. Send your my.ini file content and corresponding parts of the error log. Send the results of SHOW CREATE TABLE and SHOW TABLE STATUS for these mytable_1_old and mytable_1 tables.

I got the crashes with version 5.0.24, 5.0.27 (i see that the 5.0 w64 version is no longer present in the download page) and 5.1.x (sorry i don't remember the exact version).

The more complicated table used during the tests is the one shown below, but i got the crashes even with tables using less columns. Notice that the mytable_n_old tables have the same structure of the mytable_n tables but only the primary index

CREATE TABLE `mytable_1` (
  `ProgCollo` int(11) unsigned NOT NULL default '0',
  `NumeroLotto` int(11) unsigned default NULL,
  `NumeroLista` int(11) unsigned default NULL,
  `NumeroOrdine` int(11) unsigned default NULL,
  `Delivery` int(11) default NULL,
  `Tipo` char(3) default NULL,
  `Quantita` int(11) default NULL,
  `Volume` int(11) default NULL,
  `Peso` int(11) default NULL,
  `PesoImballo` int(11) default NULL,
  `Elaborato` tinyint(1) default NULL,
  `Dataora` datetime default NULL,
  `Trasmesso` tinyint(1) default NULL,
  `DataoraTrasmesso` datetime default NULL,
  `DataoraImballo` datetime default NULL,
  `Fattura` tinyint(1) default NULL,
  `VolumeImballo` int(11) default NULL,
  `Numerobaia` smallint(6) default NULL,
  `Flags` smallint(6) default NULL,
  `TipoImballo` char(2) default NULL,
  `Collovettore` varchar(30) default NULL,
  `Ncollo` smallint(6) default NULL,
  `Volumeplc` int(10) unsigned default NULL,
  `Cartone` varchar(10) default NULL,
  `Stationid` tinyint(3) unsigned default NULL,
  `Bancale` int(10) unsigned default NULL,
  `Gested` tinyint(3) unsigned default NULL,
  PRIMARY KEY  (`ProgCollo`),
  KEY `Ordine` (`NumeroOrdine`),
  KEY `Station` (`Stationid`),
  KEY `Bancale` (`Bancale`),
  KEY `Lista` (`NumeroLista`,`NumeroOrdine`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 

I used the INNODB engine only.
Sorry i can not post the exact my.ini and error_log because i have not them any more! 
But about my.ini i can say that i started from my-innodb-heavy-4G.ini and did little changes about directories and sizes. 2 gbytes buffer pool
No exotic feature used, no replication, no raid. 
Almost everything at "default" values.

To better explain what i did: a sequence of update queries like these

INSERT INGNORE INTO table1_old SELECT * FROM table1 WHERE field1="xxxxx"
DELETE FROM table1 WHERE field1="xxxxx"
INSERT INGNORE INTO table2_old SELECT * FROM table2 WHERE field1="xxxxx"
DELETE FROM table2 WHERE field1="xxxxx"
INSERT INGNORE INTO table3_old SELECT * FROM table3 WHERE field1="xxxxx"
DELETE FROM table3 WHERE field1="xxxxx"

Please, try to repeat with a newer version, 5.1.15-beta, for 64-bit Windows, and inform about the results.

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

Please, try to repeat with a newer version, 5.1.16 (or 5.1.17 that will be released officially soon).

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".