to only look for the mysqld binary (and english error strings) and assume the
datadir from that. Then, if that datadir turns out to not exist, startup will
fail. This avoids the behavior where mysqld_safe would go off and run a totally
different binary because the data directory had been moved (even when --datadir
was specified on the command line). (Bug #7249)
instead of DATADIR.
include/config-win.h:
Added checks for environment variable MYSQL_HOME
libmysql/Makefile.shared:
Added checks for environment variable MYSQL_HOME
mysys/Makefile.am:
Added checks for environment variable MYSQL_HOME
mysys/default.c:
Removed checks for my.cnf file in datadir, instead added checking
if my.cnf exists in environment variable MYSQL_HOME
scripts/mysqld_multi.sh:
Added checks for my.cnf file in environment variable MYSQL_HOME,
if exists.
scripts/mysqld_safe.sh:
Added checks for my.cnf file in environment variable MYSQL_HOME,
if exists.
Fix calls to my_print_defaults. Some how, --loose-verbose had been added before --no-defaults/defaults-file, which must be the first option.
NOTE: this may make some installations behave differently, if they depend on this broken behavior.
scripts/mysqld_safe.sh:
Fix calls to my_print_defaults. Some how, --loose-verbose had been added before --no-defaults/defaults-file, which must be the first option.
NOTE: this may make some installations behave differently, if they depend on this broken behavior.
Fix calls to my_print_defaults. Some how, --loose-verbose had been added before --no-defaults/defaults-file, which must be the first option.
NOTE: this may make some installations behave differently, if they depend on this broken behavior.
for Solaris test, and fix if @IS_LINUX@ test in mysqld_safe itself.
configure.in:
Portability fix for FIND_PROC setting; on Solaris (and
probably others), 'ps -p $$' inside a shell script just
returns 'sh' for command line, even though $0 contains
the filename. So, use 'ps -fp $$' in the test (it shows
the full command line, e.g., 'sh configure'). Leave the
actual FIND_PROC command as-is, since mysqld itself is
not a shell script.
scripts/mysqld_safe.sh:
Portability fix for mysqld_safe on non-Linux systems. A bogus use
of 'if' and 'test' caused non-bash shells to enter a section meant
to be run only on Linux systems.
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
bug #5001, added conditional if to test if port set, then only kill the processes
for this port, not all processes (in the case of this bug where multiple servers
are killed.)
The change to the 'grep' to make sure mysqld_safe isn't killed was per Serg's
discovery that mysqld_safe would get killed.
In my testing, in killing one of the pids for a running server, the "if test ! -f $pid_file"
was the case that evaluated as true, so in order to test, I had to comment that block
out.
scripts/mysqld_safe.sh:
bug #5001, added conditional if to test if port set, then only kill the processes
for this port, not all processes (in the case of this bug where multiple servers
are killed.)
The change to the 'grep' to make sure mysqld_safe isn't killed was per Serg's
discovery that mysqld_safe would get killed.
In my testing, in killing one of the pids for a running server, the "if test ! -f $pid_file"
was the case that evaluated as true, so in order to test, I had to comment that block
out.
bug #5001, added conditional if to test if port set, then only kill the processes
for this port, not all processes (in the case of this bug where multiple servers
are killed.)
The change to the 'grep' to make sure mysqld_safe isn't killed was per Serg's
discovery that mysqld_safe would get killed.
In my testing, in killing one of the pids for a running server, the "if test ! -f $pid_file"
was the case that evaluated as true, so in order to test, I had to comment that block
out.
scripts/mysqld_safe.sh:
don't protect mysqld from multiple --user commands - mysqld can take care of itself
sql/mysqld.cc:
warn about multiple --user parameters only if their arguments differ
Output TIMESTAMP in 4.1 format for 4.1 tables (or for TIMESTAMP(19)) (portability fix)
Fixed that INTERVAL can handle big integers. (Bug #3498)
Fixed that hostname="" works identical as hostname="%" for table/column grants (Bug #3473)
mysql-test/mysql-test-run.sh:
Fixed wrong http address (Bug #3460)
mysql-test/r/func_time.result:
Results for new test cases
mysql-test/t/func_time.test:
Added test of INTERVAL with big integers
scripts/mysqld_safe.sh:
Added html address to manual in case of error
sql/field.cc:
Output TIMESTAMP in 4.1 format for 4.1 tables (or for TIMESTAMP(19))
sql/item_timefunc.cc:
Fixed that INTERVAL can handle big integers. (Bug #3498)
sql/mysql_priv.h:
Removed not needed prototype
sql/sql_acl.cc:
Fixed that hostname="" works identical as hostname="%" for table/column grants (Bug #3473)
Moved GRANT_TABLE::GRANT_TABLE functions ou from class definition to make it possible to debug them
sql/structs.h:
Fix for long values to INTERVAL
Output TIMESTAMP in 4.1 format for 4.1 tables (or for TIMESTAMP(19)) (portability fix)
Fixed that INTERVAL can handle big integers. (Bug #3498)
Fixed that hostname="" works identical as hostname="%" for table/column grants (Bug #3473)
started via mysqld_safe (BUG#2114) Thanks to Rodrigo Marchant for the
suggestion
scripts/mysqld_safe.sh:
- fixed early removing of socket file when multiple instances of mysqld are
started via the script (BUG#2114) Thanks to Rodrigo Marchant for the
suggestion
and --defaults-file is a non-existing or non-regular file.
Bug#755
scripts/mysqld_safe.sh:
Added a warning to my_print_defaults if --verbose is given
and --defaults-file is a non-existing or non-regular file.
Bug#755
if one is running many mysqlds through mysqld_multi, for example.
Without this option, on Linux one mysqld_safe process may kill
other mysqlds as well, if started using the same binary and path.
if one is running many mysqlds through mysqld_multi, for example.
Without this option, on Linux one mysqld_safe process may kill
other mysqlds as well, if started using the same binary and path.
define a different niceness level in my.cnf (Bug #627)
scripts/mysqld_safe.sh:
- applied patch from Christian Hammers to be able to define a different
niceness level in my.cnf (Bug #627)
man/perror.1:
Auto merged
scripts/mysqld_safe.sh:
Auto merged
sql/gen_lex_hash.cc:
Auto merged
sql/ha_myisam.cc:
Auto merged
sql/item.cc:
Auto merged
sql/sql_select.cc:
Auto merged
strings/ctype-tis620.c:
Auto merged
mysql-test/r/analyse.result:
ul
mysql-test/r/group_by.result:
ul
mysql-test/r/select.result:
ul
BitKeeper/etc/ignore:
auto-union
BitKeeper/deleted/.del-delete.result:
Delete: mysql-test/r/delete.result
BitKeeper/deleted/.del-stamp-h.in~4a5d6676232516c5:
Auto merged
client/mysqlbinlog.cc:
Auto merged
include/my_sys.h:
Auto merged
libmysql/libmysql.c:
Auto merged
mysql-test/t/delete.test:
Auto merged
mysql-test/t/join.test:
Auto merged
mysql-test/t/type_datetime.test:
Auto merged
mysys/my_copy.c:
Auto merged
sql/field.h:
Auto merged
sql/opt_range.cc:
Auto merged
sql/share/polish/errmsg.txt:
Auto merged
sql/sql_rename.cc:
Auto merged
sql/sql_select.cc:
Auto merged
mysql-test/mysql-test-run.sh:
merge with 3.23.56
mysql-test/r/backup.result:
merge (needs to be updated)
mysql-test/r/join.result:
merge (needs to be updated)
mysql-test/r/type_datetime.result:
merge (needs to be updated)
mysql-test/t/backup.test:
merge with 3.23
scripts/mysqld_safe.sh:
Only use first --user option
sql/ha_myisam.cc:
Don't let BACKUP TABLE overwrite old files
sql/log_event.h:
merge
sql/mysql_priv.h:
merge
sql/mysqld.cc:
Use first --user option
sql/slave.cc:
use local version
sql/sql_repl.h:
use local version
Reverted wrong patch for mysqld_safe.sh
client/mysqldump.c:
Patch for novell.
scripts/mysqld_safe.sh:
Reverted wrong patch.
(We must delete the files first as otherwise we will get a problem if the files has the wrong permissions or if the machine went down without removing the socket file)
(On Windows --log-error is enabled by default)
mysys/my_getopt.c:
Call get_one_option() also for boolean values.
scripts/mysqld_safe.sh:
Add option --log-error
sql/log.cc:
Fix for --log-error
sql/mysql_priv.h:
Fix for --log-error
sql/mysqld.cc:
Fix for --log-error
sql/sql_udf.cc:
Lock the mysql/func table properly
innobase/btr/btr0btr.c:
Auto merged
ltconfig:
Auto merged
innobase/include/btr0btr.h:
Auto merged
innobase/log/log0log.c:
Auto merged
libmysql/Makefile.shared:
Auto merged
man/isamlog.1:
Auto merged
man/mysql.1:
Auto merged
man/mysql_zap.1:
Auto merged
man/isamchk.1:
Auto merged
man/mysqlaccess.1:
Auto merged
man/mysqladmin.1:
Auto merged
man/mysqld_multi.1:
Auto merged
man/mysqld.1:
Auto merged
man/mysqld_safe.1:
Auto merged
man/mysqldump.1:
Auto merged
man/mysqlshow.1:
Auto merged
man/replace.1:
Auto merged
myisam/mi_unique.c:
Auto merged
scripts/mysqld_safe.sh:
Auto merged
sql/net_pkg.cc:
Auto merged
sql/sql_table.cc:
Auto merged
strings/Makefile.am:
Auto merged
man/perror.1:
Auto merged
Makefile.am:
merge with 3.23
client/mysqltest.c:
merge with 3.23
innobase/btr/btr0pcur.c:
merge with 3.23 (use local file)
innobase/row/row0purge.c:
merge with 3.23 (use local file)
scripts/mysql_fix_privilege_tables.sh:
Merge with 3.23 (Apply debian patches)
sql/ha_innodb.cc:
merge with 3.23 (use local file)
sql/mysqld.cc:
Merge with 3.23 (use local file)
sql/net_serv.cc:
Merge with 3.23 (use local file)
sql/sql_db.cc:
Merge with 3.23 (use local file)
This is coded to not cause a speed impact on top level AND expressions where we don't care if an AND expression returns 0 or NULL
mysql-test/r/bdb.result:
Fix results after serges last patch
mysql-test/r/innodb.result:
Fix results after serges last patch
mysql-test/r/null.result:
Update for new AND handling of NULL
scripts/mysqld_safe.sh:
Fix 'isroot' test to work even if user is not root
sql/item.h:
Fixed that NULL and 0 returns 0 instead of NULL
sql/item_cmpfunc.cc:
Fixed that NULL and 0 returns 0 instead of NULL
sql/item_cmpfunc.h:
Fixed that NULL and 0 returns 0 instead of NULL
sql/sql_base.cc:
Fixed that NULL and 0 returns 0 instead of NULL
sql/sql_parse.cc:
Fixed that NULL and 0 returns 0 instead of NULL
sql/sql_select.cc:
Fixed that NULL and 0 returns 0 instead of NULL
sql/sql_yacc.yy:
Fixed that NULL and 0 returns 0 instead of NULL
