Bug #73438 JET replicationMixedStressTest fails for 5.5->5.6.21
Submitted: 30 Jul 2014 15:56 Modified: 9 Jan 2015 7:39
Reporter: Jorge Jordao Email Updates:
Status: Closed Impact on me:
None 
Category:Tests: Replication Severity:S7 (Test Cases)
Version:5.6.21 (advanced 64bit) OS:Any
Assigned to: CPU Architecture:Any

[30 Jul 2014 15:56] Jorge Jordao
Description:
The test executes a large set of DML statements on the master, from multiple concurrent clients. Then it issues a MASTER_POS_WAIT call on the slave.

It is failing under the following conditions
- 5.5 master -> 5.6.21 slave. No error with other version combinations, including 5.6.20 slave.
- binlog-format row or mixed. No error with statement.

The error seems to occur in the slave call to SELECT MASTER_POS_WAIT.
A few minutes after the call is issued (it's a lot of updates to process), the server dies with the following stack trace
Thread pointer: 0x7fedc8000990
Attempting backtrace. You can use the following information to find out
where mysqld died. If you see no messages after this, something went
terribly wrong...
stack_bottom = 7fede2647e60 thread_stack 0x40000
/home/jorge/mysql/5.6.21/bin/mysqld(my_print_stacktrace+0x35)[0x939405]
/home/jorge/mysql/5.6.21/bin/mysqld(handle_fatal_signal+0x3d8)[0x6a2458]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x10340)[0x7fee0b2d8340]
/home/jorge/mysql/5.6.21/bin/mysqld(_Z10unpack_rowPK14Relay_log_infoP5TABLEjPKhPK9st_bitmapPS5_PmS5_+0x30a)[0x8f598a]
/home/jorge/mysql/5.6.21/bin/mysqld(_ZN20Write_rows_log_event9write_rowEPK14Relay_log_infob+0xbe)[0x8ca8fe]
/home/jorge/mysql/5.6.21/bin/mysqld(_ZN20Write_rows_log_event11do_exec_rowEPK14Relay_log_info+0x22)[0x8cac92]
/home/jorge/mysql/5.6.21/bin/mysqld(_ZN14Rows_log_event12do_apply_rowEPK14Relay_log_info+0x2f)[0x8c2fcf]
/home/jorge/mysql/5.6.21/bin/mysqld(_ZN14Rows_log_event14do_apply_eventEPK14Relay_log_info+0xb64)[0x8cefe4]
/home/jorge/mysql/5.6.21/bin/mysqld(_ZN9Log_event11apply_eventEP14Relay_log_info+0x6d)[0x8c99ed]
/home/jorge/mysql/5.6.21/bin/mysqld(_Z26apply_event_and_update_posPP9Log_eventP3THDP14Relay_log_info+0x17b)[0x9108bb]
/home/jorge/mysql/5.6.21/bin/mysqld[0x911287]
/home/jorge/mysql/5.6.21/bin/mysqld(handle_slave_sql+0xc03)[0x912713]
/home/jorge/mysql/5.6.21/bin/mysqld(pfs_spawn_thread+0x126)[0x9b1026]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x8182)[0x7fee0b2d0182]
/lib/x86_64-linux-gnu/libc.so.6(clone+0x6d)[0x7fee09f9630d]

How to repeat:
Run JET replicationMixedStressTest with 5.5 master and 5.6.21 slave
[30 Jul 2014 16:07] Jorge Jordao
DDL statements executed at the beginning

Attachment: ddl.txt (text/plain), 955 bytes.

[30 Jul 2014 16:08] Jorge Jordao
DML statements executed, along with several other random ones, on the master from multiple concurrent clients

Attachment: dml.txt (text/plain), 1.39 KiB.

[1 Aug 2014 13:51] Jorge Jordao
The error also does *not* occur if the 5.6.21 slave is on Windows.
[9 Jan 2015 7:38] Erlend Dahl
Fixed in 5.6.22 and 5.7.6 under the heading of

Bug#19704825 TEMPORARY SLAVE TYPE CONVERSION TABLES RELEASED TO EARLY