Problem description:
mysqlhotcopy fails if a view presents in the database.
Analysis:
Before 5.5 'FLUSH TABLES <tbl_name> ... WITH READ LOCK' will able
to get lock for all tables (i.e. base tables and view tables).
In 5.5 onwards 'FLUSH TABLES <tbl_name> ... WITH READ LOCK' for
'view tables' will not work, because taking flush locks on view
tables is not valid.
Fix:
Take flush lock for 'base tables' and read lock for 'view table'
separately.
Note: most of the patch has been backported from bug#13006947's patch
- Removed files specific to compiling on OS/2
- Removed files specific to SCO Unix packaging
- Removed "libmysqld/copyright", text is included in documentation
- Removed LaTeX headers for NDB Doxygen documentation
- Removed obsolete NDB files
- Removed "mkisofs" binaries
- Removed the "cvs2cl.pl" script
- Changed a few GPL texts to use "program" instead of "library"
Conflicts:
Text conflict in configure.in
Text conflict in dbug/dbug.c
Text conflict in mysql-test/r/ps.result
Text conflict in mysql-test/t/ps.test
Text conflict in sql/CMakeLists.txt
Text conflict in sql/ha_ndbcluster.cc
Text conflict in sql/mysqld.cc
Text conflict in sql/sql_plugin.cc
Text conflict in sql/sql_table.cc
Since 5.5 FLUSH TABLES requires unlocked tables so the old LOCK TABLES/FLUSH TABLES
model fails. Use atomic FLUSH TABLES WITH READ LOCK instead.
per-file comments:
scripts/mysqlhotcopy.sh
Bug#42465 mysqlhotcopy fails with "Table '<y>' was locked with a READ lock" error message.
Use FLUSH TABLES WITH READ LOCK instead of LOCK TABLES / FLUSH TABLES pair.
Keep the old scheme for older servers.
- Make sure mysqlhotcopy doesn't create unrestorable slaves.
scripts/mysqlhotcopy.sh:
Bug #27101 mysqlhotcopy --record_log_pos retrieves wrong slave parameters
- Record what's been executed by the slave, not what's been delivered.
Remove code from mysqlhotcopy which deals with the so-called "RAID"
feature of older MyISAM.
scripts/mysqlhotcopy.sh:
Remove code which tried to handle the so-called "RAID" files.
The "RAID" feature has been dropped from MySQL builds, and this
code caused mysqlhotcopy to skip (silently) over databases with
two-character names (e.g., "ab").
Better Windows support in the scripts directory
mysql_config.pl.in, mysql_install_db.pl.in:
New Perl version of Unix shell script, mainly for Windows
Many files in scripts directory:
Use default Perl location "#!/usr/bin/perl" instead of the build host path
extra/CMakeLists.txt:
Added target for executable "resolveip"
extra/resolveip.c:
Exclude Unix specific headers when compiling on Windows
scripts/CMakeLists.txt:
On Windows, filter Perl scripts and change name from ".sh" to ".pl"
mysql_convert_table_format.sh mysql_explain_log.sh
mysql_secure_installation.sh mysql_tableinfo.sh
mysqld_multi.sh mysqldumpslow.sh
mysqlhotcopy.sh
Do the same for the new Windows specific Perl versions of shell scripts
mysql_config.pl.in mysql_install_db.pl.in
In CMake, set reasonable values for 'CFLAGS', 'prefix', 'datadir' and so on.
scripts/Makefile.am:
Include "mysql_config.pl.in" and "mysql_install_db.pl.in" in the source TAR
scripts/make_win_bin_dist:
Only include explicitly listed scripts from the "scripts" directory
scripts/mysql_convert_table_format.sh:
Use default Perl location "#!/usr/bin/perl" instead of the build host path
scripts/mysql_explain_log.sh:
Use default Perl location "#!/usr/bin/perl" instead of the build host path
scripts/mysql_tableinfo.sh:
Use default Perl location "#!/usr/bin/perl" instead of the build host path
scripts/mysqld_multi.sh:
Use default Perl location "#!/usr/bin/perl" instead of the build host path
scripts/mysqldumpslow.sh:
Use default Perl location "#!/usr/bin/perl" instead of the build host path
scripts/mysqlhotcopy.sh:
Use default Perl location "#!/usr/bin/perl" instead of the build host path
scripts/mysql_config.pl.in:
New Perl version of Unix shell script, mainly for Windows
scripts/mysql_install_db.pl.in:
New Perl version of Unix shell script, mainly for Windows
Use "SHOW TABLES FROM `db`" instead of $dbh->tables() in the
get_list_of_tables() routine.
The symptom is that, when used with recent versions of DBD::mysql,
mysqlhotcopy uses a double-qualified table name, for example:
Invalid db.table name 'test.test`.`x' at /usr/bin/mysqlhotcopy line 855.
This is caused by a change in DBD::mysql. See this diff:
http://svn.perl.org/viewcvs/modules/DBD-mysql/trunk/lib/DBD/mysql.pm?r1=9183&r2=9188
Basically, older DBD::mysql implemented a limited ->table_info method;
now the full method is implemented, and as a result DBI's ->tables()
method has access to the schema value, so it uses it.
scripts/mysqlhotcopy.sh:
Use "SHOW TABLES FROM `db`" instead of $dbh->tables() in
get_list_of_tables(). DBI's ->tables() method calls
"SHOW TABLES" under the hood, and then quotes the
names and (depending on DBD::mysql version) adds the schema
name, too. mysqlhotcopy doesn't want the schema name
or the quotes, so it's all around better to just call
"SHOW TABLES" itself.
scripts/mysqlhotcopy.sh:
Defined new raid directory regular expression variable that allows a mix of two alphanumeric characters. Replaced hardcode raid directory regular expressions with new variable.
BitKeeper/etc/ignore:
Added scripts/mysqlhotcopy.sh.rej to the ignore list
scripts/mysqlhotcopy.sh:
BUG#7967 Use fetchrow_hashref() when parsing SHOW SLAVE STATUS so
that --record-log-pos option will work with all versions from 3.23
forward and will not break again in the future if additional columns
are added.
DBD::mysql that quote table names are used by removing the quoting. Also fix
a bug that stripped all '~' characters out of the table regex instead of
just the leading one used to negate the expression. (Bug #8136)
scripts/mysqlhotcopy.sh:
remove quotes around table names
BitKeeper/etc/logging_ok:
auto-union
innobase/include/sync0arr.h:
Auto merged
innobase/sync/sync0arr.c:
Auto merged
scripts/mysqlhotcopy.sh:
Auto merged
innobase/srv/srv0srv.c:
SCCS merged
Noteworthy:
- New HANDLER code
- New multi-update-grant-check code
- Table lock code in ha_innodb.cc was not applied
BitKeeper/etc/logging_ok:
auto-union
BitKeeper/deleted/.del-ctype-latin1_de.c~c5d8f9208bceb98e:
Auto merged
Build-tools/mysql-copyright-2:
Auto merged
acinclude.m4:
Auto merged
client/mysqladmin.c:
Auto merged
client/mysqldump.c:
Auto merged
include/config-win.h:
Auto merged
include/my_global.h:
Auto merged
include/myisam.h:
Auto merged
innobase/btr/btr0btr.c:
Auto merged
innobase/buf/buf0buf.c:
Auto merged
ltmain.sh:
Auto merged
innobase/dict/dict0dict.c:
Auto merged
innobase/fsp/fsp0fsp.c:
Auto merged
innobase/include/dict0dict.h:
Auto merged
innobase/include/row0mysql.h:
Auto merged
innobase/log/log0log.c:
Auto merged
innobase/log/log0recv.c:
Auto merged
innobase/pars/pars0opt.c:
Auto merged
innobase/row/row0row.c:
Auto merged
innobase/sync/sync0arr.c:
Auto merged
innobase/ut/ut0dbg.c:
Auto merged
myisam/mi_check.c:
Auto merged
myisam/mi_close.c:
Auto merged
myisam/mi_create.c:
Auto merged
myisam/mi_locking.c:
Auto merged
myisam/myisampack.c:
Auto merged
mysql-test/r/delete.result:
Auto merged
mysql-test/r/func_if.result:
Auto merged
Build-tools/mysql-copyright:
Merge with 4.0 (too most of the code from 4.0)
Makefile.am:
merge
client/mysql.cc:
Used 4.1 code
configure.in:
merge
innobase/os/os0file.c:
merge
innobase/row/row0mysql.c:
merge
mysql-test/r/ctype_latin1_de.result:
merge
mysql-test/r/flush_table.result:
merge
mysql-test/r/func_str.result:
merge
mysql-test/r/handler.result:
merge
mysql-test/r/multi_update.result:
merge
mysql-test/r/type_timestamp.result:
Removed testing of 'new' mode, as this is only relevant for 4.0
mysql-test/r/update.result:
merge
mysql-test/t/delete.test:
merge
mysql-test/t/flush_table.test:
merge
mysql-test/t/func_str.test:
merge
mysql-test/t/handler.test:
merge
mysql-test/t/multi_update.test:
merge
mysql-test/t/type_timestamp.test:
Removed testing of 'new' mode, as this is only relevant for 4.0
mysql-test/t/update.test:
merge
mysys/errors.c:
merge
mysys/my_fstream.c:
merge
mysys/my_pread.c:
merge
mysys/my_write.c:
merge
mysys/mysys_priv.h:
merge
scripts/mysqlhotcopy.sh:
merge
sql/field.cc:
Keep code from 4.1
sql/field.h:
Keep code from 4.1
sql/ha_innodb.cc:
Don't merge lock code from 4.0; Heikki will look at this
sql/ha_myisam.cc:
merge
sql/handler.cc:
merge
sql/item_cmpfunc.cc:
merge
sql/item_cmpfunc.h:
merge
sql/item_strfunc.cc:
merge
sql/mysql_priv.h:
merge
sql/mysqld.cc:
merge
sql/protocol.cc:
merge
sql/records.cc:
merge
sql/repl_failsafe.cc:
merge
mysql-test/r/lock_multi.result:
merge
mysql-test/t/ctype_latin1_de.test:
merge
mysql-test/t/func_if.test:
merge
mysql-test/t/lock_multi.test:
merge
sql/repl_failsafe.h:
merge
Remove unnessessary header protection
sql/slave.h:
merge
sql/sql_acl.cc:
merge
sql/sql_base.cc:
merge
sql/sql_cache.cc:
auto merge
sql/sql_class.cc:
merge
sql/sql_class.h:
merge
sql/sql_delete.cc:
merge
sql/sql_handler.cc:
Get new HANDLER code into 4.1
sql/sql_parse.cc:
Keep old file
sql/sql_repl.cc:
merge
sql/sql_repl.h:
merge
sql/sql_show.cc:
merge
sql/sql_table.cc:
merge
sql/sql_union.cc:
Applied the examine_rows bug fix from 4.0 by hand
sql/sql_update.cc:
New multi-update-grant-check code from 4.0
sql/sql_yacc.yy:
New multi-update-grant-check code from 4.0
sql/stacktrace.c:
merge
sql/table.h:
merge
when copying a small index file because the value returned
for $length is < 1024. This can happen if the filehandle
was open()ed as an UTF-8 encoded file with Unicode characters
(In this case read() returns characters not bytes)
(Thanks to Mike Bethune) for this hint)
Build-tools/Do-compile:
Fixed indentation
configure.in:
Added patches from Novell
Added C_EXTRA_FLAGS as an easy way to pass flags to both CFLAGS and CXXFLAGS
extra/perror.c:
Fixed error number reporting to not report 'Unknown error'
include/my_global.h:
Defines to make NETWARE patches cleaner
include/thr_alarm.h:
Fixed wrong macro
netware/mysql_install_db.c:
Indentation fix
"mysqlhotcopy fails to copy tables but does not indicate a failure"
("does not indicate a failure");
this is about "mysqlhotcopy fails to copy tables".
mysql-test/t/rpl_error_ignored_table-slave.opt:
added a missing newline
scripts/mysqlhotcopy.sh:
Fix for BUG#812.
The problem was that with many tables to copy (10000 in the bug's example),
the generated 'cp' command line was 1MB long, whereas (at least on my Linux) it
should not exceed 128 kB. Testing the 'cp' in a shell terminal gives
"arguments list too long".
So we issue several small (100 kB) 'cp' command lines instead of a big one.
Of course, this will still fail on systems where the limit is below 100 kB.
We now have safe_system() which cuts the command line in pieces,
and calls safe_simple_system() (execution) for each piece.
Undo serg's patch that used my_print_defaults,
and instead simply do not define $opt{user}.
This allows the script to use option files properly.
scripts/mysqlhotcopy.sh:
Undo serg's patch that used my_print_defaults,
and instead simply do not define $opt{user}.
This allows the script to use option files properly.
the --password parameter (better add it to my.cnf instead) (Thanks to
Christian Hammers for the suggestion)
scripts/mysqlhotcopy.sh:
- added word of warning about providing the password as cleartext using
the --password parameter (better add it to my.cnf instead)
myisam/mi_open.c:
Fixed problem with wrongly calculated max_data_file_length
mysql-test/Makefile.am:
Added missing .require test files
scripts/mysqlhotcopy.sh:
Remove end / from directory names (portability fix)
tests/grant.res:
Update of test results
Added option --host to mysqlhotcopy
mysql-test/mysql-test-run.sh:
Added error message if the server doesn't start.
Increase connect timeout a bit (for running under purify).
mysql-test/t/rpl000001.test:
Longer sleep time (for running under purify)
scripts/mysqlhotcopy.sh:
Added option --host for usage with TCP/IP connections
sql/gen_lex_hash.cc:
Fixed typo
include/sslopt-usage.h:
Paul's patch for --ssl-xxx options for the --help message
scripts/mysqlhotcopy.sh:
Paul's patch for --port option
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
