MySQL Bugs: #18266: Mysqldump issue continually dropping connection

Bug #18266	Mysqldump issue continually dropping connection
Submitted:	15 Mar 2006 23:16	Modified:	27 Jun 2006 14:26
Reporter:	Doug Dalton	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Server	Severity:	S3 (Non-critical)
Version:	4.1.13	OS:	Linux (Suse 10.x smp /Freebsd 6.0)
Assigned to:		CPU Architecture:	Any

Description:
I think this is a server issue but I am not sure,  I am running 
mysql  Ver 14.7 Distrib 4.1.13, for suse-linux (x86_64) using readline 5.0
mysqld  Ver 4.1.13-log for suse-linux on x86_64 (SUSE MySQL RPM)
mysqldump  Ver 10.9 Distrib 4.1.13, for suse-linux (x86_64)

I issue:

mysqldump -X -u xxx -pxxx "--where=XXX=4" db table >output.sql

mysqldump: Error 2013: Lost connection to MySQL server during query when dumping table `table` at row: 9972148
*** glibc detected *** free(): invalid next size (fast): 0x0000000000623200 ***
 23969 Aborted                 mysqldump -X -u xxx -pxxx "--where=XXX=4" db table >output.sql

This aborts regardless of a localhost or remote connection (direct machine to machine)

There are no errors in the mysql error log and the server continues to run

I have set my connectio timeouts super high to stop this error to no avail:

interactive_timeout=1000000
wait_timeout=10000000
connect_timeout=100

How to repeat:
mysqldump -X -u xxx -pxxx "--where=XXX=4" db table >output.sql

on a db with 20 million records

Suggested fix:
Unknown

Thank you for a problem report. Can you try to install the newer version of MySQL server and tools, 4.1.18, and check the same command? Have you seen similar problem on a smaller table, with other options to mysqldump? Please, send also the SHOW TABLE STATUS results for the table you are dumping.

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".

MySQL 4.1.12 on Mandriva Linux 2006:

I have several big mnoGoSearch databases (great indexer/search engine, the same as yours ;-)) 

I have made a script to dump each database individually in its own file :
--------------------------
MYSQLDUMP_COMMAND="mysqldump  --max_allowed_packet=100000000 --protocol=socket --single-transaction   --password=passwd"
DATE2=`date "+%Y-%m-%d-%Hh%Mm%Ss"`

 mysql --password=passwd -e "show databases"| while read X; do 
        echo "Backuping $X"; 
        RESULT=`$MYSQLDUMP_COMMAND "$X"| bzip2 -c > "$BACKUP_DIR/MySQl-dump-$X-$DATE2.SQL.bz2" 2>&1`; 
        if [ -n "$RESULT" ]; then
                echo "Error with $X: $RESULT"
        fi
        done
------------------------------

It works fine for all databases but the biggest one:
 - the whole db is 10 GO 
 - 2 tables are 3,5 and 4,5 GO
 - one table have more than 1 975 000 entries

Error message: 

mysqldump: Error 2013: Lost connection to MySQL server during query when dumping table bdict at row 369947.

FYI: bdict table is 3,3 GO and have only 844000 entries.
The bzip2 process slows down the dumping process greatly, and the .bz2 file is between 1GO and 2 GO when it stops with this error..

So have an idea?

All reporters:

Please, try to repeat with a newer version, 4.1.19, 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".