Bug#407406: mysql-server: DELETE FROM table causes Incorrect key file for table

2007-01-23 Thread Alexis Sukrieh

Christian Hammers wrote:

Hello Alexis


Hi Christian


Maybe your innodb data files were corrupted somewhen else and the problem
only just appeared at this occasion. If you try to reproduce it in a second
database on your host, e.g. in a chroot so it uses a different inndb table
space file, do you get the same error?


I'm going to try that, but FYI, after having downgraded mysql-server to 
the sarge version (4.x) the bug disappeared (using the same dump files).



Until we proved that this is a general problem I downgraded the bugreport
as it would prevent 5.0.32 from entering testing/etch. But please try to
reproduce it again on your computer, maybe I just did something different.


I understand, I'm going to try reproduce it in a chroot with etch's 
mysql-server.


By the way, I'm wondering if that would be a good idea to provide both 
version 4 and 5 within etch instead of making the version 5 provide 
the 4. Don't you think?


Regards,


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#407406: mysql-server: DELETE FROM table causes Incorrect key file for table

2007-01-22 Thread Christian Hammers
tags 407406 + unreproducible
severity normal
stop

Hello Alexis

On 2007-01-18 Alexis Sukrieh wrote:
 Christian Hammers wrote:
  That bug is supposed to be fixed since 5.0.26 if I read it correctly.
 
 Yep, but I'm facing a symptom very near to the upstream bug reported.
 That's why I thought in the first place it was the same bug.

I was unable to reproduce the bug with your SQL files. I created a
chroot with Debian sarge, there installed MySQL and loaded it with your
SQL files and afterwards upgraded the chroot to Debian testing/etch.
Both of your queries worked fine in sarge as well as in testing/etch.

Maybe your innodb data files were corrupted somewhen else and the problem
only just appeared at this occasion. If you try to reproduce it in a second
database on your host, e.g. in a chroot so it uses a different inndb table
space file, do you get the same error?

Until we proved that this is a general problem I downgraded the bugreport
as it would prevent 5.0.32 from entering testing/etch. But please try to
reproduce it again on your computer, maybe I just did something different.

bye,

-christian-


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#407406: mysql-server: DELETE FROM table causes Incorrect key file for table

2007-01-18 Thread Christian Hammers
tags 407406 moreinfo
stop

Hello

On 2007-01-18 Alexis Sukrieh wrote:
 Package: mysql-server
 Version: 5.0.30-3
 Severity: grave
 Justification: causes serious data loss
 
 
 Hello, 
 
 I use a Debian etch server for development purposes, and I use there a
 mysql database.
 
 Now that I've upgraded my development server from sarge to etch, I face
 a grave bug: when importing data from a dump made by mysqldump to a new
 database, a key corruption appears.
 
 Here is the error message:
 
 Mysql::Error: #HY000Incorrect key file for table '/tmp/#sql_7b82_0.MYI';
 try to repair it: query ...
 
 After Googling a bit, I found that bug on the MySQL's bug tracking system:
 http://bugs.mysql.com/bug.php?id=22384

That bug is supposed to be fixed since 5.0.26 if I read it correctly.

Could you mail me a copy of the mysqldump output you tried to insert? 
Having something reproducible vastly increases the chance to fix the bug.

bye,

-christian-



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#407406: mysql-server: DELETE FROM table causes Incorrect key file for table

2007-01-18 Thread Alexis Sukrieh

Christian Hammers wrote:

That bug is supposed to be fixed since 5.0.26 if I read it correctly.


Yep, but I'm facing a symptom very near to the upstream bug reported.
That's why I thought in the first place it was the same bug.

Could you mail me a copy of the mysqldump output you tried to insert? 
Having something reproducible vastly increases the chance to fix the bug.


I can mail you the dump but only privately (and GPG encrypted) as it's 
some private data, used for a project I'm working on for a company.


I ask you not to distribute the dump even if that's development stuff.

Thanks a lot for paying attention to that issue.

Here is how to reproduce the bug:
$ echo create database test_database | mysql
$ mysql test_database  db/create.sql
$ mysql test_database  db/data.sql
$ mysql test_database
[...]
mysql select words.id, words.value, sum(count) as count from sitetags, 
words where words.id = word_id group by word_id order by count desc 
limit 100;ERROR 126 (HY000): Incorrect key file for table 
'/tmp/#sql_7b82_0.MYI'; try to repair it

mysql

I'm mailing you the dump (it's quite big).

Regards,

Alexis


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#407406: mysql-server: DELETE FROM table causes Incorrect key file for table

2007-01-18 Thread Alexis Sukrieh

Hi again,

I also have another bug (certainly related) with the same dump:

mysql SHOW FIELDS FROM users;
ERROR 1030 (HY000): Got error 28 from storage engine
mysql


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#407406: mysql-server: DELETE FROM table causes Incorrect key file for table

2007-01-17 Thread Alexis Sukrieh
Package: mysql-server
Version: 5.0.30-3
Severity: grave
Justification: causes serious data loss


Hello, 

I use a Debian etch server for development purposes, and I use there a
mysql database.

Now that I've upgraded my development server from sarge to etch, I face
a grave bug: when importing data from a dump made by mysqldump to a new
database, a key corruption appears.

Here is the error message:

Mysql::Error: #HY000Incorrect key file for table '/tmp/#sql_7b82_0.MYI';
try to repair it: query ...

After Googling a bit, I found that bug on the MySQL's bug tracking system:
http://bugs.mysql.com/bug.php?id=22384

So it sounds to be fixed upstream (the report speaks about a patch).

IMHO, That bug should definitely not be shipped with etch, it renders any
impacted database unusable.

Regards,

-- System Information:
Debian Release: 4.0
  APT prefers testing
  APT policy: (500, 'testing')
Architecture: i386 (i686)
Shell:  /bin/sh linked to /bin/bash
Kernel: Linux 2.6.16-2-vserver-k7
Locale: [EMAIL PROTECTED], [EMAIL PROTECTED] (charmap=UTF-8) (ignored: LC_ALL 
set to fr_FR.UTF-8)

Versions of packages mysql-server depends on:
ii  mysql-server-5.0  5.0.30-3   mysql database server binaries

mysql-server recommends no packages.

-- debconf information:
  mysql-server/nis_warning:
  mysql-server/really_downgrade_from_41: false
  mysql-server/mysql_update_hints1:
  mysql-server/start_on_boot: true
  mysql-server/postrm_remove_databases: false
* mysql-server/mysql_install_db_notes:


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]