SHOW PROCESSLIST, SHOW BINLOGS
Problem: A deadlock was occurring when 4 threads were
involved in acquiring locks in the following way
Thread 1: Dump thread ( Slave is reconnecting, so on
Master, a new dump thread is trying kill
zombie dump threads. It acquired thread's
LOCK_thd_data and it is about to acquire
mysys_var->current_mutex ( which LOCK_log)
Thread 2: Application thread is executing show binlogs and
acquired LOCK_log and it is about to acquire
LOCK_index.
Thread 3: Application thread is executing Purge binary logs
and acquired LOCK_index and it is about to
acquire LOCK_thread_count.
Thread 4: Application thread is executing show processlist
and acquired LOCK_thread_count and it is
about to acquire zombie dump thread's
LOCK_thd_data.
Deadlock Cycle:
Thread 1 -> Thread 2 -> Thread 3-> Thread 4 ->Thread 1
The same above deadlock was observed even when thread 4 is
executing 'SELECT * FROM information_schema.processlist' command and
acquired LOCK_thread_count and it is about to acquire zombie
dump thread's LOCK_thd_data.
Analysis:
There are four locks involved in the deadlock. LOCK_log,
LOCK_thread_count, LOCK_index and LOCK_thd_data.
LOCK_log, LOCK_thread_count, LOCK_index are global mutexes
where as LOCK_thd_data is local to a thread.
We can divide these four locks in two groups.
Group 1 consists of LOCK_log and LOCK_index and the order
should be LOCK_log followed by LOCK_index.
Group 2 consists of other two mutexes
LOCK_thread_count, LOCK_thd_data and the order should
be LOCK_thread_count followed by LOCK_thd_data.
Unfortunately, there is no specific predefined lock order defined
to follow in the MySQL system when it comes to locks across these
two groups. In the above problematic example,
there is no problem in the way we are acquiring the locks
if you see each thread individually.
But If you combine all 4 threads, they end up in a deadlock.
Fix:
Since everything seems to be fine in the way threads are taking locks,
In this patch We are changing the duration of the locks in Thread 4
to break the deadlock. i.e., before the patch, Thread 4
('show processlist' command) mysqld_list_processes()
function acquires LOCK_thread_count for the complete duration
of the function and it also acquires/releases
each thread's LOCK_thd_data.
LOCK_thread_count is used to protect addition and
deletion of threads in global threads list. While show
process list is looping through all the existing threads,
it will be a problem if a thread is exited but there is no problem
if a new thread is added to the system. Hence a new mutex is
introduced "LOCK_thd_remove" which will protect deletion
of a thread from global threads list. All threads which are
getting exited should acquire LOCK_thd_remove
followed by LOCK_thread_count. (It should take LOCK_thread_count
also because other places of the code still thinks that exit thread
is protected with LOCK_thread_count. In this fix, we are changing
only 'show process list' query logic )
(Eg: unlink_thd logic will be protected with
LOCK_thd_remove).
Logic of mysqld_list_processes(or file_schema_processlist)
will now be protected with 'LOCK_thd_remove' instead of
'LOCK_thread_count'.
Now the new locking order after this patch is:
LOCK_thd_remove -> LOCK_thd_data -> LOCK_log ->
LOCK_index -> LOCK_thread_count
INTO CLIENT_ERRORS ARRAY
Analysis:
--------
The client may crash while executing a statement due to
the missing mapping of the server error to it's equivalent
client error.
When trying to reallocate memory for the packet buffer, if
the system is out of memory or the packet buffer is large,
the server errors 'ER_OUT_OF_RESOURCES' or 'ER_PACKET_TOO_LARGE'
is returned respectively. The client error number calculated is
negative and when trying to dereference the array of client
error messages with the calculated error number, the client
crashes.
Fix:
----
Map the server error returned to it's equivalent client error
prior to dereferencing the array of client error messages.
Note: Test case is not added since it is difficult to simulate
the error condition.
INTO CLIENT_ERRORS ARRAY
Analysis:
--------
The client may crash while executing a statement due to
the missing mapping of the server error to it's equivalent
client error.
When trying to reallocate memory for the packet buffer, if
the system is out of memory or the packet buffer is large,
the server errors 'ER_OUT_OF_RESOURCES' or 'ER_PACKET_TOO_LARGE'
is returned respectively. The client error number calculated is
negative and when trying to dereference the array of client
error messages with the calculated error number, the client
crashes.
Fix:
----
Map the server error returned to it's equivalent client error
prior to dereferencing the array of client error messages.
Note: Test case is not added since it is difficult to simulate
the error condition.
The reason was that a couple of variables that hold number of rows that was used to calculate buffers was uint and caused an overflow.
Fixed by changing variables that could hold number of rows from uint to ulong and also added a cast for this test.
include/heap.h:
Reorder to get better alignment. Changed variables that could hold number of rows from uint to ulong
mysql-test/suite/heap/heap.result:
Added test case
mysql-test/suite/heap/heap.test:
Added test case
mysql-test/suite/plugins/t/server_audit.test:
Added sleep as we want to have disconnect logged before we try a new connect
storage/heap/ha_heap.cc:
Changed variables that could hold number of rows from uint to ulong
Limit number of rows to 4G (as most of the variables that holds rows are ulong anyway)
reset records_changed when key_stat_version is changed to not cause increments for every row changed
storage/heap/ha_heap.h:
changed records_changed to ulong as this can get big
storage/heap/hp_create.c:
Changed variables that could hold number of rows from uint to ulong
Added cast (fixed the original bug)
storage/heap/hp_delete.c:
Changed variables that could hold number of rows from uint to ulong
storage/heap/hp_open.c:
Removed not needed cast
storage/heap/hp_write.c:
Changed variables that could hold number of rows from uint to ulong
support-files/compiler_warnings.supp:
Removed extra : from supression
Due to how gap locks work, two transactions could group commit together on the
master, but get lock conflicts and then deadlock due to different thread
scheduling order on slave.
For now, remove these deadlocks by running the parallel slave in READ
COMMITTED mode. And let InnoDB/XtraDB allow statement-based binlogging for the
parallel slave in READ COMMITTED.
We are also investigating a different solution long-term, which is based on
relaxing the gap locks only between the transactions running in parallel for
one slave, but not against possibly external transactions.
Automatic merge, except for server_audit.cc that had to be modified slightly
Changes to xtradb and innobase where ignored was these made no sence for 10.0
Remove memory warnings if mysql client aborts early
Changed copyright for clients
client/mysql.cc:
Free memory if get_options fails, so that we don't get warnings from safemalloc
include/welcome_copyright_notice.h:
Added SkySQL to client copyrights
mysql-test/valgrind.supp:
Added suppressions for memory leaks from dlopen() for OpenSUSE 12.3
storage/oqgraph/mysql-test/oqgraph/regression_mdev5744.result:
Suppress warning
storage/oqgraph/mysql-test/oqgraph/regression_mdev5744.test:
Suppress warning
my_atomic_load() is implemented as __sync_fetch_and_or(var, 0) which
writes or-ed value back to var. Memory writes as such have worse
performance and scalability than reads.
gcc 4.7 and up offers better facility for atomic loads/stores. Use it
whenever it is available.
The issue was that create...trigger part of the test suite used a debug_sync point that before was never triggered (in other words, wrong meaningless test).
With the new create ... replace code the debug sync point is triggered and the test case could not handled that.
I fixed this by adding a wait and go for the debug syncpoint in the test.
Removed some compiler warnings from mysql_cond_timedwait
include/mysql/psi/mysql_thread.h:
Removed compiler warnings
mysql-test/r/create-big.result:
New test result
mysql-test/t/create-big.test:
Fixed test case as create_table_select_before_check_if_exists was not before triggered by the code.
Reduced number of my_hash_sort_bin() calls from 4 to 1 per query.
Reduced number of memory accesses done by my_hash_sort_bin().
Details:
- let MDL subsystem use pre-calculated hash value for hash
inserts and deletes
- let table cache use pre-calculated MDL hash value
- MDL namespace is excluded from hash value calculation, so that
hash value can be used by table cache as is
- hash value for MDL is calculated as resulting hash value + MDL
namespace
- extended hash implementation to accept user defined hash function
This is port of fix for MySQL BUG#17647863.
revno: 5572
revision-id: jon.hauglid@oracle.com-20131030232243-b0pw98oy72uka2sj
committer: Jon Olav Hauglid <jon.hauglid@oracle.com>
timestamp: Thu 2013-10-31 00:22:43 +0100
message:
Bug#17647863: MYSQL DOES NOT COMPILE ON OSX 10.9 GM
Rename test() macro to MY_TEST() to avoid conflict with libc++.
MDEV-4984: Implement MASTER_GTID_WAIT() and @@LAST_GTID.
MDEV-4726: Race in mysql-test/suite/rpl/t/rpl_gtid_stop_start.test
MDEV-5636: Deadlock in RESET MASTER
MASTER_GTID_WAIT() is similar to MASTER_POS_WAIT(), but works with a
GTID position rather than an old-style filename/offset.
@@LAST_GTID gives the GTID assigned to the last transaction written
into the binlog.
Together, the two can be used by applications to obtain the GTID of
an update on the master, and then do a MASTER_GTID_WAIT() for that
position on any read slave where it is important to get results that
are caught up with the master at least to the point of the update.
The implementation of MASTER_GTID_WAIT() is implemented in a way
that tries to minimise the performance impact on the SQL threads,
even in the presense of many waiters on single GTID positions (as
from @@LAST_GTID).
which has been removed earlier, but somehow got into the sources again.
This file is not really needed, and it prints warnings
"sys/timeb.h is deprecated" for every file when compiling on FreeBSD.
Cleanups:
* remove unused HA_EXTRA_EXPORT (can be added later if needed, e.g. for Aria)
* clarify the meaning of HA_CAN_EXPORT
* make all engines that support EXPORT to announce it
* reduce code duplication
Update InnoDB to 5.6.14
Apply MySQL-5.6 hack for MySQL Bug#16434374
Move Aria-only HA_RTREE_INDEX from my_base.h to maria_def.h (breaks an assert in InnoDB)
Fix InnoDB memory leak
mysql_upgrade --help now also prints out --default options and variable values.
mysql_upgrade now prints permission errors.
mysql_upgrade doesn't print some non essential info if --silent is used.
Added handler error message about incompatible versions
Fixed that mysqlbug and mysql_install_db have the executable flag set.
Removed executable flag for some non executable files.
Changed in mysql_install_db askmonty.org to mariadb.com.
Ensured that all client executables prints --default options the same way.
Allow REPAIR ... USE_FRM for old .frm files if the are still compatible.
Extended shown error for storage engine messages.
client/mysql.cc:
print_defaults() should be first (as in all other programs)
client/mysql_upgrade.c:
--help now also prints out --default options and variable values
Print out error if wrong permissions
Don't print info if --silent
client/mysqladmin.cc:
print_defaults() should be first (as in all other programs)
client/mysqlbinlog.cc:
Added print_defaults() to --help
client/mysqlcheck.c:
Added empty line in --help
client/mysqlimport.c:
Added empty line in --help
client/mysqlshow.c:
Made --help compatible
client/mysqlslap.c:
Made --help compatible
client/mysqltest.cc:
Added print_defaults() to --help
include/handler_ername.h:
Added handler error message
include/my_base.h:
Added handler error message
mysql-test/r/mysql_upgrade.result:
Updated results
mysql-test/r/repair.result:
Added test case for better error messages
mysql-test/std_data/host_old.MYD:
Added test case for better error messages
mysql-test/std_data/host_old.MYI:
Added test case for better error messages
mysql-test/std_data/host_old.frm:
Added test case for better error messages
mysql-test/t/repair.test:
Added test case for better error messages
mysys/my_handler_errors.h:
Added handler error message
scripts/CMakeLists.txt:
Fixed that mysqlbug and mysql_install_db have the executable flag set
scripts/mysql_install_db.sh:
askmonty.org -> mariadb.com
sql/ha_partition.cc:
Sometimes table_type() can be called for errors even if partition didn't manage to open any files
sql/handler.cc:
Write clear text for not handled, but defined error messages.
sql/share/errmsg-utf8.txt:
Extended shown error for storage engine messages
sql/sql_admin.cc:
Allow REPAIR ... USE_FRM for old .frm files if the are still compatible
storage/myisam/ha_myisam.cc:
Use new error message
Description: A typo in create_tailoring() causes the "contraction_flags" to be written
into cs->contractions in the wrong place. This causes two problems:
(1) Anyone relying on `contraction_flags` to decide "could this character be
part of a contraction" is 100% broken.
(2) Anyone relying on `contractions` to determine the weight of a contraction
is mostly broken
Analysis: When we are preparing the contraction in create_tailoring(), we are corrupting the
cs->contractions memory location which is supposed to store the weights(8k) + contraction information(256 bytes). We started storing the contraction information after the 4k location. This is because of logic flaw in the code.
Fix: When we create the contractions, we need to calculate the contraction with (char*) (cs->contractions + 0x40*0x40) from ((char*) cs->contractions) + 0x40*0x40. This makes the "cs->contractions" to move to 8k bytes and stores the contraction information from there. Similarly when we are calculating it for like range queries we need to calculate it from the 8k bytes onwards, this can be done by changing the logic to (const char*) (cs->contractions + 0x40*0x40). And for ucs2 charsets we need to modify the my_cs_can_be_contraction_head() and my_cs_can_be_contraction_tail() to point to 8k+ locations.
Description: A typo in create_tailoring() causes the "contraction_flags" to be written
into cs->contractions in the wrong place. This causes two problems:
(1) Anyone relying on `contraction_flags` to decide "could this character be
part of a contraction" is 100% broken.
(2) Anyone relying on `contractions` to determine the weight of a contraction
is mostly broken
Analysis: When we are preparing the contraction in create_tailoring(), we are corrupting the
cs->contractions memory location which is supposed to store the weights(8k) + contraction information(256 bytes). We started storing the contraction information after the 4k location. This is because of logic flaw in the code.
Fix: When we create the contractions, we need to calculate the contraction with (char*) (cs->contractions + 0x40*0x40) from ((char*) cs->contractions) + 0x40*0x40. This makes the "cs->contractions" to move to 8k bytes and stores the contraction information from there. Similarly when we are calculating it for like range queries we need to calculate it from the 8k bytes onwards, this can be done by changing the logic to (const char*) (cs->contractions + 0x40*0x40). And for ucs2 charsets we need to modify the my_cs_can_be_contraction_head() and my_cs_can_be_contraction_tail() to point to 8k+ locations.