Rename innodb_table_locks_old_behavior -> innodb_table_locks
Set innodb_table_locks to off by default to get same behaviour as in MySQL 4.0.20
(This means that Innodb ignore table locks by default, which makes it easier to combine MyISAM and InnoDB to simulate a transaction)
Simple optimzations and cleanups
Removed compiler warnings and fixed portability issues
Added client functions 'mysql_embedded()' to allow client to check if we are using embedded server
Fixes for purify
Change error code to HA_ERR_ROW_IS_REFERENCED if we cannot DROP a parent table referenced by a FOREIGN KEY constraint; this error number is less misleading than the previous value HA_ERR_CANNOT_ADD_FOREIGN, but misleading still; we should introduce to 5.0 a proper MySQL error code
Fix bug #5137: if innodb_file_per_table was specified, CREATE TEMPORARY TABLE ... TYPE=InnoDB said that cannot find path specified, and made mysqld to exit(1)
Raise maximum column prefix len to 767 bytes, so that MySQL can create a column prefix index of 255 UTF-8 characters (each takes 3 bytes at the maximum); add comments about why innobase_get_at_most_n_mbchars() works ok
dict0mem.h:
Raise maximum column prefix len to 767 bytes, so that MySQL can create a column prefix index of 255 UTF-8 characters (each takes 3 bytes at the maximum)
row0mysql.c:
If MySQL tries to create a column prefix index longer that 255 UTF-8 characters, give an error, and drop the table from the InnoDB internal data dictionary. MySQL did not drop the table there in its own error handling.
A partial bugfix to a multibyte charset / column prefix index bug: my_charpos() does not handle right some cases, we try to mask the bug for ASCII chars < 128 in the UTF-8 charset
column types TIMESTAMP is NOT NULL by default, so in order to have
TIMESTAMP column holding NULL valaues you have to specify NULL as
one of its attributes (this needed for backward compatibility).
Main changes:
Replaced TABLE::timestamp_default_now/on_update_now members with
TABLE::timestamp_auto_set_type flag which is used everywhere
for determining if we should auto-set value of TIMESTAMP field
during this operation or not. We are also use Field_timestamp::set_time()
instead of handler::update_timestamp() in handlers.
This allows one to setup a master <-> master replication with non conflicting auto-increment series.
Cleaned up binary log code to make it easyer to add new state variables.
Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave).
Simplified binary log handling.
Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
Improve the comment on stored_select_lock_type
ha_innodb.cc:
Let InnoDB remember select_lock_type inside LOCK TABLES, also over plain consistent read SELECTs; fix Bug #5538 : assertion failure when using mysqldump with the -l option; in MERGING this patch to 4.1, there may be PROBLEMS; that is because previous patch was never merged to 4.1; Heikki Tuuri has to polish the code in 4.1 after this patch has been merged.
* Renamed handler::estimate_number_of_rows to handler::estimate_rows_upper_bound function, which can also return "unknown"
* made filesort to use full sort buffer if number of rows to sort is not known.
If ALTER TABLE ... DROP FOREIGN KEY ... fails because of a wrong constraint name, return a table handler error number 150 instead of 152; the value 152 was misleading, as it referred to '152 = Cannot delete a parent row', whereas '150 = Foreign key constraint is incorrectly formed' is less misleading
innobase_mysql_tmpfile(): call dup() and my_close() on the file
returned by create_temp_file()
in order to avoid memory leak caused by my_open() being paired with close()
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
mysqld.cc, ha_innodb.cc:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
- client side part is simple and may be considered stable
- server side part now just joggles with THD state to save execution
state and has no additional locking wisdom.
Lot's of it are to be rewritten.
next-key locking. Using this option the locks InnoDB sets on index
records do not affect the ``gap'' before that index record. Thus, this option
allows phantom problem.
more logical table/index_flags
return HA_ERR_WRONG_COMMAND instead of abstract methods where appropriate
max_keys and other limits renamed to max_supported_keys/etc
max_keys/etc are now wrappers to max_supported_keys/etc
ha_index_init/ha_rnd_init/ha_index_end/ha_rnd_end are now wrappers to real {index,rnd}_{init,end} to enforce strict pairing
Won't be pushed as is - separate email sent for internal review.
WL#1717 "binlog-innodb consistency".
Now when mysqld starts, if InnoDB does a crash recovery, we use the binlog name
and position retrieved from InnoDB (corresponding to the last transaction
successfully committed by InnoDB) to cut any rolled back transaction from
the binary log. This is triggered by the --innodb-safe-binlog option.
Provided you configure mysqld to fsync() InnoDB at every commit (using
flush_log_at_trx_commit) and to fsync() the binlog at every write
(using --sync-binlog=1), this behaviour guarantees that a master always has
consistency between binlog and InnoDB, whenever the crash happens.
6 tests to verify that it works.
/***********************************************************************
This function stores binlog offset and flushes logs */
void innobase_store_binlog_offset_and_flush_log(char *binlog_name,longlong offset)
requested by Guilhem to ha_innodb.cc and ha_innodb.h.
Change made by Jan.Lindstrom@innodb.com
Ensured that all projects compile
Removed compiler warnings
Better setting of server_version variable.
Fix that make_win_src_distribution creates the privilege tables.
New records_in_range() interface (similar to read_range())
Macros for faster bitmap handling
Simplify read_range() code (#WL1786)
New general key_cmp() function to compare keys
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
Reserve the MySQL LOCK_thread_count mutex when printing thd->query of
an arbitrary transaction; if we are printing thd->query of a transaction that
we know is currently executing inside InnoDB, then we know that MySQL cannot
meanwhile change thd->query, and no need to reserve the MySQL mutex; note
that this patch still leaves open the possibility of races in MySQL's
thd->query_len
A flawed fix of the thd->query race in SHOW INNODB STATUS; see the comments in code about how to fix this properly; we cannot use LOCK_thread_count to protect thd->query, because that will cause a deadlock of threads
Fixed bugs in group_concat with ORDER BY and DISTINCT (Bugs #2695, #3381 and #3319)
Fixed crash when doing rollback in slave and the io thread catched up with the sql thread
Set locked_in_memory properly
Final version of patch.
Adds support for specifying of DEFAULT NOW() and/or ON UPDATE NOW()
clauses for TIMESTAMP field definition.
Current implementation allows only one such field per table and
uses several unireg types for storing info about this properties of
field. It should be replaced with better implementation when new
.frm format is introduced.
"(binlog, position) stored by InnoDB for a replication slave can be wrong".
This code contains conditional #if to distinguish between versions;
it should be merged into 4.1 and 5.0.
Change some inline functions to normal functions, so that they can be called in ha_innodb.cc; this removes the the remaining linking problems reported by Paul DuBois
Multiple charset support for InnoDB; note that if you create tables and the default charset-collation is not latin1_swedish_ci, then you cannot use those tables if you downgrade to 4.0 again
Make the drop InnoDB database operation aware of the setting FOREIGN_KEY_CHECKS=0; DROP DATABASE seems to work ok anyway, but that is because MySQL first calls DROP TABLE for each table which has an .frm file
Allow always DROPping of a table which is only referenced by FOREIGN KEY constraints from the same table
Many files:
Do not let REPLACE to perform internally an UPDATE if the table is referenced by a FOREIGN KEY: the manual says that REPLACE must resolve a duplicate key error semantically with DELETE(s) + INSERT, and not by an UPDATE; the internal update caused foreign key checks and cascaded operations to behave in a semantically wrong way
If AUTOCOMMIT=1, then we do not need to make a plain SELECT set shared locks even on the SERIALIZABLE isolation level, because we know the transaction is read-only: a read-only transaction can always be performed on the REPEATABLE READ level, and that does not endanger the serializability
Remove redundant code; parse both the database name and the table name in a FOREIGN KEY constraint with quotes in mind
row0mysql.c, ha_innodb.cc, sql_table.cc:
Return error message Cannot delete or update a parent row... if we try to drop a table which is referenced by a FOREIGN KEY constraint, and the user has not set foreign_key_checks=0
Added support for lower_case_table_names=2, which is to be used on case insensitive file systems.
This tells MySQL to preserve the used case of filenames and database names to make it esier to move files between cases sensitive can case insensitive file systems (like Windows and Linux)
Cleaned up embedded library access and query cache handling
Changed min stack size to 128K (to allow longer MyISAM keys)
Fixed wrong priority for XOR (should be less than NEG to get -1^1 to work)
This is the main commit for Worklog tasks:
* A more dynamic binlog format which allows small changes (1064)
* Log session variables in Query_log_event (1063)
Below 5.0 means 5.0.0.
MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed),
SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think
of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this
works for queries, except LOAD DATA INFILE (for this it would have to wait
for Dmitri's push of WL#874, which in turns waits for the present push, so...
the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1,
5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1.
Apart from that, the new binlog format is designed so that it can tolerate
a little variation in the events (so that a 5.0.0 slave could replicate a
5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I
later add replication of charsets it should break nothing. And when I later
add a UID to every event, it should break nothing.
The main change brought by this patch is a new type of event, Format_description_log_event,
which describes some lengthes in other event types. This event is needed for
the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event,
we can later add more bytes to the header of every event without breaking compatibility.
Inside Query_log_event, we have some additional dynamic format, as every Query_log_event
can have a different number of status variables, stored as pairs (code, value); that's
how SQL_MODE and session variables and catalog are stored. Like this, we can later
add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows
if we want.
MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs.
Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs),
upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs);
so both can be "hot" upgrades.
Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0.
3.23 and 4.x can't be slaves of 5.0.
So downgrading from 5.0 to 4.x may be complicated.
Log_event::log_pos is now the position of the end of the event, which is
more useful than the position of the beginning. We take care about compatibility
with <5.0 (in which log_pos is the beginning).
I added a short test for replication of SQL_MODE and some other variables.
TODO:
- after committing this, merge the latest 5.0 into it
- fix all tests
- update the manual with upgrade notes.
Fixed problem with char > 128 in QUOTE() function. (Bug #1868)
Disable creation of symlinks if my_disable_symlink is set
Fixed searching of TEXT with end space. (Bug #1651)
Fixed caching bug in multi-table-update where same table was used twice. (Bug #1711)
Fixed problem with UNIX_TIMESTAMP() for timestamps close to 0. (Bug #1998)
Fixed timestamp.test
Fixed compiler warnings (a lot of hidden variables detected by the Forte compiler)
Added a lot of 'version_xxx' strings to 'show variables'
Prevent copying of TMP_TABLE_PARAM (This caused core dump bug on Solaris)
Fixed problem with printing sub selects to debug log
Added option --max-record-length=# to myisamchk
Don't try repair twice if doing myisamchk --repair --force
Shared memory handler didn't clean up things on errors or shutdown
Fix the OS error 2 reported by Miguel and Mark in Windows crash recovery: a * had been forgotten from the path in directory scanning
ha_innodb.cc:
Set default directory in fil0fil.c right if we are running the Embedded Server Library, where the default dir of the process is not necessarily the MySQL datadir
Multiple tablespaces for InnoDB
sql_table.cc:
Tell explicitly that InnoDB should retrieve all columns in CHECKSUM TABLE
sql_update.cc, sql_select.cc, my_base.h:
More descriptive flag name HA_EXTRA_RETRIEVE_ALL_COLS
Fix for a bug with LEAST() in WHERE clause
ha_innodb.cc:
Fix for a configure bug
multi_update.result, multi_update.test:
Fix for the update with NULL's in the result set
sql_update.cc:
Fix for the update with NULL's in the result set
Fix for the update with NULL's in the result set
Fix for the update with NULL's in the result set
Since the mysql interactive client or some other phase in communication cuts > 64 kB strings to very short, restrict the output of SHOW INNODB STATUS to at most 60 kB
Remove potential starvation of a full log buffer flush: only flush up to the lsn which was the largest at the time when we requested the full log buffer flush
os0sync.h, os0sync.c:
Fix a bug in os_event on Unix: even though we signaled the event, some threads could continue waiting if the event became nonsignaled quickly again; this made group commit less efficient than it should be
In ORDER BY MySQL seems to set the key read flag also in the case where the primary key contains only a prefix of a column - not the whole column; to prevent potential bugs retrieve the whole column if the index contains a prefix of it
Cleanup
ha_innodb.cc, data0type.h:
Make sure non-latin1 users can downgrade from 4.0.14 to an earlier version if they have not created DATA_BLOB column prefix indexes
Merge InnoDB-4.0.14: SAVEPOINT now implemented; InnoDB now accepts also column prefix keys; crashing bug in ON UPDATE CASCADE fixed; page checksum formula fixed
Remove accidenatlly committed debug printfs when query cache is used
sql_update.cc, sql_load.cc, sql_insert.cc, sql_delete.cc:
For the transactional query cache algorithm to work we must invalidate the query cache in INSERT/DELETE/UPDATE before writing to the binlog or calling ha_autocommit_... Note that binlog writing may also call commit. The crucial thing is that the transaction which modified data must not be committed BEFORE the query cache is invalidated.
Remove code never actually used in IGNORE; InnoDB just rolls back the latest row insert or update on a duplicate key error and leaves it for MySQL to decide whether to ignore the error
Make ANALYZE TABLE to update index cardinality and other statistics also for an InnoDB table; note that the estimates are based on 10 random dives, not on an exhaustive read of the table
Prevent crash if MySQL calls rnd_end WITOUT calling rnd_init first: only restore active_index if it the old value was pushed in rnd_init
Initialize active_index to a sensible value: prevent crash if we restore it to the original value after an table scan
Make InnoDB to restore old active_index value after a table scan: MySQL may assume that a scan does NOT change active_index; this partially fixes bug 241 of UPDATE ... ORDER BY ... but it still remains that MySQL actually ignores the ORDER BY for both MyISAM and InnoDB tables
log0recv.c:
Use fflush to make sure report of a corrupt log record is printed to .err log before mysqld crashes
Fix bug: MySQL could erroneously return Empty set if InnoDB estimated index range size to 0 records though the range was not empty; MySQL also failed to do the next-key locking in the case of an empty index range
Remove (already commented out) code about storing the latest relay log pos info to InnoDB: we will probably use only the master log pos info when we restart a slave after a crash
Prepare for 5.x where HA_READ_PREFIX_LAST may pass only a few first bytes of the last field in a key value
ha_innodb.cc:
In 4.0 always assume HA_READ_PREFIX_LAST passes a complete-field prefix of a key value; LIKE queries use a padding trick
Fix potential bug: if MySQL calls store_lock with the TL_IGNORE argument in the middle of query processing, then InnoDB select_lock_type could be reset to LOCK_NONE in a wrong place
Call pthread_mutex_destroy() on not used mutex.
Changed comments in .h and .c files from // -> /* */
Added detection of mutex on which one didn't call pthread_mutex_destroy()
Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression"
Added optimisation for ORDER BY NULL
Temporary fix to make the new client/server protocol to work: fetch all columns in an index, even if not needed by the SELECT statement; Monty has to fix the parser to tell InnoDB what to fetch
Test allocation of memory beforehand if we are trying to return a > 2 MB BLOB; normally InnoDB asserts if memory allocation fails
ha_innodb.cc:
Do not fetch all columns if change_active_index() is called during a query; a sum(a), max(a) query seemed to do that, doing unnecessary copying (the change actually made in the previous bk ci)
Free BLOB heap of handle when MySQL calls some ::extra()'s
Add diagnostic prints to determine why the 'queries inside InnoDB' might drift upwards
ha_innodb.cc:
Add more print space to SHOW INNODB STATUS; remove possible memory leak in case there is an error
Changed long packat handling to check for packets of length 0xffffff.
This does however break packet handling for older clients.
If you are using packets >= 16M then you need to upgrade client and server
after this patch.
Class for sending data from server to client (Protocol)
This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
Make copies of all string-type start/up options in case C-sharp moves them around; remove the need to specify innodb_log_arch_dir in my.cnf, since it has no relevance anyway
Add support for LIMIT # OFFSET #
Changed lock handling: Now all locks should be stored in TABLE_LIST instead of passed to functions.
Don't call query_cache_invalidate() twice in some cases
mysql_change_user() now clears states to be equal to close + connect.
Fixed a bug with multi-table-update and multi-table-delete when used with LOCK TABLES
Fixed a bug with replicate-do and UPDATE
