MySQL Bugs: #39286: Deletes on partition tables takes large time on slave to execute using row based

Bug #39286	Deletes on partition tables takes large time on slave to execute using row based
Submitted:	6 Sep 2008 3:17	Modified:	8 Oct 2008 6:41
Reporter:	bhushan uparkar	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Server: Row Based Replication ( RBR )	Severity:	S2 (Serious)
Version:	5.1.25	OS:	Any (Debian)
Assigned to:		CPU Architecture:	Any
Tags:	dead slow execution, deletes, partitioning, row based replication

Description:
Hello,

I am using MySql 5.1.25 on debian linux,

dbbackup@dmdb902:~$ uname -a
Linux dmdb902 2.6.18-5-amd64 #1 SMP Thu May 31 23:51:05 UTC 2007 x86_64 GNU/Linux

Server: Dell 2950 with 2 Quad processor 2.33 GHz
Ram: 32 GB

 We are using Master/Slave setup with row based replication. When we delete the records from a partition based tables on master it executes within fraction of seconds [ in this case I deleted 44K records ], but on slave it took 6+ hours to execute the same delete opration. This causes the slave to be out of sync for entire duration. Though I was able to do various queries on slave, its responding fine. Following are some of the status from innodb engine related to above "delete operation on slave",
=======================
dbbackup@dmdb902:~$ uname -a
Linux dmdb902 2.6.18-5-amd64 #1 SMP Thu May 31 23:51:05 UTC 2007 x86_64 GNU/Linux
dbbackup@dmdb902:~$ mysql -u dbbackup -e "show engine innodb status \G" | grep deleted
Number of rows inserted 6943572056, updated 2984030186, deleted 10281730, read 383560006516
dbbackup@dmdb902:~$ mysql -u dbbackup -e "show engine innodb status \G" | grep deleted
Number of rows inserted 6943572056, updated 2984030186, deleted 10282622, read 383560606931
dbbackup@dmdb902:~$ uname -a
Linux dmdb902 2.6.18-5-amd64 #1 SMP Thu May 31 23:51:05 UTC 2007 x86_64 GNU/Linux
dbbackup@dmdb902:~$ mysql -u dbbackup -e "show engine innodb status \G" | grep deleted
Number of rows inserted 6943572056, updated 2984030186, deleted 10283413, read 383560607722

============================

 The slave executes the replication log very very slow.  In the past we have deleted the records from partitioned tables , and observed the same problem. But this problem did not happened quite consistently.
Let me know if you have question or need more details on this problem.

How to repeat:
1] Setup master and slave database using row based replication with build 5.1.25
2] Create partitioned table using range based partition,
3] Populate each partition with 100k -125k records per partition
4] Populate minimum 90 partitions
5] Delete the 40K-50K record from one of the partition
6] You will observe the slave for execution of delete

------------

Partition table information

Attachment: fact_table.rtf (application/rtf, text), 26.92 KiB.

Thank you for a problem report. Please, try to repeat with a newer version, 5.1.26, on both master and slave and inform about the results. In case of the same problem, please, send my.cnf from master and slave.

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".