Bug #5550 Server crash in test 'rpl000001'
Submitted: 13 Sep 2004 19:24 Modified: 27 Jun 2005 15:09
Reporter: Joerg Bruehe Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Server Severity:S3 (Non-critical)
Version:4.1.5 OS:Linux (Linux 2.4.21 IA64 SMP)
Assigned to: Lenz Grimmer CPU Architecture:Any

[13 Sep 2004 19:24] Joerg Bruehe
Description:
Running the test suite on the ICC-generated binaries leads to a server crash in 'rpl000001':

 TEST                           RESULT
------------------------------------------
rpl000001                      [ fail ]

Errors are (from /home/mysqldev/quadita2-icc-glibc23/mysql-4.1.5-gamma/mysql-test/var/log/mysqltest-time) :
/home/mysqldev/quadita2-icc-glibc23/mysql-4.1.5-gamma/client/mysqltest: At line 15: query 'stop slave' failed: 2013: Lost connection to MySQL server during query
(the last lines may be the most important ones)

Last lines of './var/log/slave.trace':
T@23058430: | | | >end_io_cache
T@23058430: | | | | >my_free
T@23058430: | | | | | my: ptr: 0x6000000000ad9d20
T@23058430: | | | | <my_free
T@23058430: | | | <end_io_cache
T@23058430: | | | >my_free
T@23058430: | | | | my: ptr: 0x6000000000a83c30
T@23058430: | | | <my_free
T@23058430: | | | >my_free
T@23058430: | | | | my: ptr: 0x6000000000a83c90
T@23058430: | | | <my_free
T@23058430: | | <close_cached_file
T@23058430: | | info: freeing host
T@23058430: | | >free_root
T@23058430: | | | enter: root: 0x6000000000ad8a58  flags: 0
T@23058430: | | <free_root
T@23058430: | | >free_root
T@23058430: | | | enter: ro  ((followed by lots of hex 0x00 bytes))

No core file produced.

How to repeat:
Run the test (suite).
[14 Sep 2004 14:58] Lenz Grimmer
BUG#5334 was marked as a duplicate of this bug.
[11 Oct 2004 14:45] Joerg Bruehe
4.1.6-gamma still shows this error! Machine 'quadita2', Intel compiler "icc".
[12 Oct 2004 12:27] Dmitry Lenev
Simpler test case for this bug in mysqltest format:

-------------------------------------------
source include/master-slave.inc;

save_master_pos;
connection slave;
sync_with_master;
stop slave;
--sleep 5
select 1;
-------------------------------------------

Additional testing using non-optimized build showed that bug probably caused by error which pops up during/right after execution of STOP SLAVE statement. Runtime error occurs somewhere at the end of handle_slave_io (and handle_slave_sql) function. More precise it pops up after last statement of this function (during stack unwinding/thread termination ?) and causes slave abortion.

Error log for slave reflects this:
-------------------------------------------
CURRENT_TEST: rpl_min
/home/mysqldev/dlenev/mysql-4.1.6-gamma/sql/mysqld: ready for connections.
Version: '4.1.6-gamma-standard-debug-log'  socket: '/home/mysqldev/dlenev/mysql-4.1.6-gamma/mysql-test/var/tmp/slave.sock'  port: 9309  Official MySQL-standard binary
041012 14:51:55  [NOTE] Slave SQL thread initialized, starting replication in log 'FIRST' at position 0, relay log '/home/mysqldev/dlenev/mysql-4.1.6-gamma/mysql-test/var/log/slave-relay-bin.000001' position: 4
041012 14:51:55  [NOTE] Slave I/O thread: connected to master 'root@127.0.0.1:9306',  replication started in log 'FIRST' at position 4
041012 14:51:55  [ERROR] next log '/home/mysqldev/dlenev/mysql-4.1.6-gamma/mysql-test/var/log/slave-relay-bin.000002' is currently active
041012 14:51:55  [ERROR] Slave I/O thread killed while reading event
041012 14:51:55  [ERROR] Slave I/O thread exiting, read up to log 'master-bin.000001', position 79
041012 14:51:55  [ERROR] Error reading relay log event: slave SQL thread was killed

ERROR: Incorrect Phase

ERROR: Incorrect Phase
-----------------------------------------
(Two last errors are errors from run-time and not MySQL errors...)

Altough this can be stack corruption, it also may be a compiler bug.
I will contact ICC team about these issue (at least it would be nice to know what exactly these error message mean)...
[19 Oct 2004 21:07] Lenz Grimmer
Still happens with latest Intel icc compiler version:

Intel(R) C++ Itanium(R) Compiler for Itanium(R)-based applications
Version 8.1    Build 20040922 Package ID: l_cc_pc_8.1.022
[27 Jun 2005 15:09] Lenz Grimmer
Bug does no longer occur with icc Version "l_cc_pc_8.1.032" - Closing.