Bug #45977 rpl_backup_shutdown fails on pb-1 due to result mismatch and warning in valgrind
Submitted: 6 Jul 2009 16:15 Modified: 24 Aug 2009 9:20
Reporter: Alfranio Junior Email Updates:
Status: Duplicate Impact on me:
None 
Category:MySQL Server: Backup Severity:S3 (Non-critical)
Version:6.0 OS:Any
Assigned to: Assigned Account CPU Architecture:Any

[6 Jul 2009 16:15] Alfranio Junior
Description:
rpl.rpl_backup_shutdown fails due to result mismatch:

@@ -114,6 +114,9 @@
 RESTORE FROM 'db12m.bak';
 backup_id
 #
+Warnings:
+#	1194	Table 'backup_history' is marked as crashed and should be repaired
+#	1194	Table 'backup_progress' is marked as crashed and should be repaired
 # Slave will be stopped as result of restore incident event
 # Connecting to slave...
 # Check the slave status

mysqltest: Result content mismatch

 - saving '/dev/shm/var-rpl_binlog_row-112/log/rpl.rpl_backup_shutdown/' to '/dev/shm/var-rpl_binlog_row-112/log/rpl.rpl_backup_shutdown/'
 - found 'core.14538' (0/5)

In valgring, we have what follows:

==6362== 
==6362== Conditional jump or move depends on uninitialised value(s)
==6362==    at 0x4CB6BBA: _db_enter_ (dbug.c:1761)
==6362==    by 0x4C9C9F2: my_malloc (my_malloc.c:29)
==6362==    by 0x4CA5927: init_dynamic_string (string.c:39)
==6362==    by 0x414ECF: run_query(st_connection*, st_command*, int) (mysqltest.cc:7095)
==6362==    by 0x416736: main (mysqltest.cc:7880)

How to repeat:
info: http://tinyurl.com/ksc6hb http://tinyurl.com/nkuylv 
[platforms: any, row-mix]

xref: http://tinyurl.com/nf6tab
[4 Aug 2009 14:12] Chuck Bell
Revised Problem Statement
-------------------------
The problem with the backup logs is the server is crashing on the --shutdown 10 command on line #60 in the test. When the server crashes while logs are open it can cause these warnings to appear when the server is restarted.

We need to fix the shutdown issue first before diagnosing anything wrong with the backup subsystem. The backup logs will be fine if the server is shutdown properly. 

As a workaround, it is suggested using a different mechanism to shutdown the server like mysqladmin.

Reassigning to replication team for investigation into the shutdown crash.

Note: BUG#45238 may have a causal relationship -- whatever is causing that bug may be related.

Note: Removing the 'not_windows' in master-slave.inc will allow the developer to see this problem on Windows where it is consistent and reproduces the original problem statement for the bug.
[24 Aug 2009 9:20] Alfranio Junior
This is a duplicate of BUG#46013.

revno: 1810.3925.14
revision-id: davi.arnaut@sun.com-20090813200720-utqy73cj0orcy80z