log_buf_pool_get_oldest_modification(): Acquire
log_sys_t::flush_order_mutex in order to prevent a race condition
that was introduced in
commit 1a6f708ec5 (MDEV-15058).
Before that change, log_buf_pool_get_oldest_modification()
was protected by both log_sys.mutex and log_sys.flush_order_mutex
like it was supposed to be ever since
commit a52c4820a3 (MySQL 5.5.10).
buf_pool_t::get_oldest_modification(): Replaces
buf_pool_get_oldest_modification(), to emphasize that
log_sys.flush_order_mutex must be acquired by the caller if needed.
log_close(): Invoke log_buf_pool_get_oldest_modification()
in order to ensure a clean shutdown.
The scenario of the race condition is as follows:
1. The buffer pool is clean (no writes are pending).
2. mtr_add_dirtied_pages_to_flush_list() releases log_sys.mutex.
3. log_buf_pool_get_oldest_modification() observes that the
buffer pool is clean and returns log_sys.lsn.
4. log_checkpoint() completes, writing a wrong checkpoint header
according to which everything up to log_sys.lsn was clean.
5. mtr_add_dirtied_pages_to_flush_list() completes the execution
of mtr_memo_note_modifications(), releases the page latches and
the flush_order_mutex.
6. On a subsequent log_checkpoint(), the assertion could fail
if the page modifications had not been flushed yet.
The failing assertion (which is valid) was added in MySQL 5.7
mysql/mysql-server@5c6c6ec693
and merged to MariaDB Server 10.2.2 in
commit fec844aca8.
Thanks to MDEV-15058, there is only one InnoDB buffer pool.
Allocating buf_pool statically removes one level of pointer indirection
and makes code more readable, and removes the awkward initialization of
some buf_pool members.
While doing this, we will also declare some buf_pool_t data members
private and replace some functions with member functions. This is
mostly affecting buffer pool resizing.
This is not aiming to be a complete rewrite of buf_pool_t to
a proper class. Most of the buffer pool interface, such as
buf_page_get_gen(), will remain in the C programming style
for now.
buf_pool_t::withdrawing: Replaces buf_pool_withdrawing.
buf_pool_t::withdraw_clock_: Replaces buf_withdraw_clock.
buf_pool_t::create(): Repalces buf_pool_init().
buf_pool_t::close(): Replaces buf_pool_free().
buf_bool_t::will_be_withdrawn(): Replaces buf_block_will_be_withdrawn(),
buf_frame_will_be_withdrawn().
buf_pool_t::clear_hash_index(): Replaces buf_pool_clear_hash_index().
buf_pool_t::get_n_pages(): Replaces buf_pool_get_n_pages().
buf_pool_t::validate(): Replaces buf_validate().
buf_pool_t::print(): Replaces buf_print().
buf_pool_t::block_from_ahi(): Replaces buf_block_from_ahi().
buf_pool_t::is_block_field(): Replaces buf_pointer_is_block_field().
buf_pool_t::is_block_mutex(): Replaces buf_pool_is_block_mutex().
buf_pool_t::is_block_lock(): Replaces buf_pool_is_block_lock().
buf_pool_t::is_obsolete(): Replaces buf_pool_is_obsolete().
buf_pool_t::io_buf: Make default-constructible.
buf_pool_t::io_buf::create(): Delayed 'constructor'
buf_pool_t::io_buf::close(): Early 'destructor'
HazardPointer: Make default-constructible. Define all member functions
inline, also for derived classes.
Some fields were protected by log_sys.mutex, which adds quite some
overhead for readers. Some readers were submitting dirty reads.
log_t::lsn: Declare private and atomic. Add wrappers get_lsn()
and set_lsn() that will use relaxed memory access. Many accesses
to log_sys.lsn are still protected by log_sys.mutex; we avoid the
mutex for some readers.
log_t::flushed_to_disk_lsn: Declare private and atomic, and move
to the same cache line with log_t::lsn.
log_t::buf_free: Declare as size_t, and move to the same cache line
with log_t::lsn.
log_t::check_flush_or_checkpoint_: Declare private and atomic,
and move to the same cache line with log_t::lsn.
log_get_lsn(): Define as an alias of log_sys.get_lsn().
log_get_lsn_nowait(), log_peek_lsn(): Remove.
log_get_flush_lsn(): Define as an alias of log_sys.get_flush_lsn().
log_t::initiate_write(): Replaces log_buffer_sync_in_background().
Compute MONITOR_LSN_CHECKPOINT_AGE on demand in
srv_mon_process_existing_counter().
This allows us to remove the overhead of MONITOR_SET
calls for the counter.
Introduce special synchronization primitive group_commit_lock
for more efficient synchronization of redo log writing and flushing.
The goal is to reduce CPU consumption on log_write_up_to, to reduce
the spurious wakeups, and improve the throughput in write-intensive
benchmarks.
Our benchmarking efforts indicate that the reasons for splitting the
buf_pool in commit c18084f71b
have mostly gone away, possibly as a result of
mysql/mysql-server@ce6109ebfd
or similar work.
Only in one write-heavy benchmark where the working set size is
ten times the buffer pool size, the buf_pool->mutex would be
less contended with 4 buffer pool instances than with 1 instance,
in buf_page_io_complete(). That contention could be alleviated
further by making more use of std::atomic and by splitting
buf_pool_t::mutex further (MDEV-15053).
We will deprecate and ignore the following parameters:
innodb_buffer_pool_instances
innodb_page_cleaners
There will be only one buffer pool and one page cleaner task.
In a number of INFORMATION_SCHEMA views, columns that indicated
the buffer pool instance will be removed:
information_schema.innodb_buffer_page.pool_id
information_schema.innodb_buffer_page_lru.pool_id
information_schema.innodb_buffer_pool_stats.pool_id
information_schema.innodb_cmpmem.buffer_pool_instance
information_schema.innodb_cmpmem_reset.buffer_pool_instance
Redo log subsystem was decoupled from tablespace subsystem. It now manages file
descriptors for redo log files by itself.
FIL_TYPE_LOG: removed, code in various places was simplified
SRV_LOG_SPACE_FIRST_ID: renamed to SRV_SPACE_ID_UPPER_BOUND
to better match its purpose. Code in various places was simplified
fil_n_log_flushes: replaced with log_sys::flushes
fil_n_pending_log_flushes: replaced with log_sys::pending_flushes
log_t::files::files: redo log file descriptors
log_t::files::file_names: redo log file names
log_t::files::set_file_names(): set file names without opening them
log_t::files::open_files(): opens redo log files
log_t::files::read(): treats several files as one big
log_t::files::write(): treats several files as one big
log_t::files::fsync(): flushes page cache to disk
log_t::files::close_files(): closes redo log files
fil_open_log_and_system_tablespace_files(): renamed to
fil_open_system_tablespace_files()
and obviously it now doesn't open redo log files
global files[1000]: removed. Why it was needed at all?
We will remove the InnoDB background operation of merging buffered
changes to secondary index leaf pages. Changes will only be merged as a
result of an operation that accesses a secondary index leaf page,
such as a SQL statement that performs a lookup via that index,
or is modifying the index. Also ROLLBACK and some background operations,
such as purging the history of committed transactions, or computing
index cardinality statistics, can cause change buffer merge.
Encryption key rotation will not perform change buffer merge.
The motivation of this change is to simplify the I/O logic and to
allow crash recovery to happen in the background (MDEV-14481).
We also hope that this will reduce the number of "mystery" crashes
due to corrupted data. Because change buffer merge will typically
take place as a result of executing SQL statements, there should be
a clearer connection between the crash and the SQL statements that
were executed when the server crashed.
In many cases, a slight performance improvement was observed.
This is joint work with Thirunarayanan Balathandayuthapani
and was tested by Axel Schwenke and Matthias Leich.
The InnoDB monitor counter innodb_ibuf_merge_usec will be removed.
On slow shutdown (innodb_fast_shutdown=0), we will continue to
merge all buffered changes (and purge all undo log history).
Two InnoDB configuration parameters will be changed as follows:
innodb_disable_background_merge: Removed.
This parameter existed only in debug builds.
All change buffer merges will use synchronous reads.
innodb_force_recovery will be changed as follows:
* innodb_force_recovery=4 will be the same as innodb_force_recovery=3
(the change buffer merge cannot be disabled; it can only happen as
a result of an operation that accesses a secondary index leaf page).
The option used to be capable of corrupting secondary index leaf pages.
Now that capability is removed, and innodb_force_recovery=4 becomes 'safe'.
* innodb_force_recovery=5 (which essentially hard-wires
SET GLOBAL TRANSACTION ISOLATION LEVEL READ UNCOMMITTED)
becomes safe to use. Bogus data can be returned to SQL, but
persistent InnoDB data files will not be corrupted further.
* innodb_force_recovery=6 (ignore the redo log files)
will be the only option that can potentially cause
persistent corruption of InnoDB data files.
Code changes:
buf_page_t::ibuf_exist: New flag, to indicate whether buffered
changes exist for a buffer pool page. Pages with pending changes
can be returned by buf_page_get_gen(). Previously, the changes
were always merged inside buf_page_get_gen() if needed.
ibuf_page_exists(const buf_page_t&): Check if a buffered changes
exist for an X-latched or read-fixed page.
buf_page_get_gen(): Add the parameter allow_ibuf_merge=false.
All callers that know that they may be accessing a secondary index
leaf page must pass this parameter as allow_ibuf_merge=true,
unless it does not matter for that caller whether all buffered
changes have been applied. Assert that whenever allow_ibuf_merge
holds, the page actually is a leaf page. Attempt change buffer
merge only to secondary B-tree index leaf pages.
btr_block_get(): Add parameter 'bool merge'.
All callers of btr_block_get() should know whether the page could be
a secondary index leaf page. If it is not, we should avoid consulting
the change buffer bitmap to even consider a merge. This is the main
interface to requesting index pages from the buffer pool.
ibuf_merge_or_delete_for_page(), recv_recover_page(): Replace
buf_page_get_known_nowait() with much simpler logic, because
it is now guaranteed that that the block is x-latched or read-fixed.
mlog_init_t::mark_ibuf_exist(): Renamed from mlog_init_t::ibuf_merge().
On crash recovery, we will no longer merge any buffered changes
for the pages that we read into the buffer pool during the last batch
of applying log records.
buf_page_get_gen_known_nowait(), BUF_MAKE_YOUNG, BUF_KEEP_OLD: Remove.
btr_search_guess_on_hash(): Merge buf_page_get_gen_known_nowait()
to its only remaining caller.
buf_page_make_young_if_needed(): Define as an inline function.
Add the parameter buf_pool.
buf_page_peek_if_young(), buf_page_peek_if_too_old(): Add the
parameter buf_pool.
fil_space_validate_for_mtr_commit(): Remove a bogus comment
about background merge of the change buffer.
btr_cur_open_at_rnd_pos_func(), btr_cur_search_to_nth_level_func(),
btr_cur_open_at_index_side_func(): Use narrower data types and scopes.
ibuf_read_merge_pages(): Replaces buf_read_ibuf_merge_pages().
Merge the change buffer by invoking buf_page_get_gen().
This patch contains a full implementation of the optimization
that allows to use in-memory rowid / primary filters built for range
conditions over indexes. In many cases usage of such filters reduce
the number of disk seeks spent for fetching table rows.
In this implementation the choice of what possible filter to be applied
(if any) is made purely on cost-based considerations.
This implementation re-achitectured the partial implementation of
the feature pushed by Galina Shalygina in the commit
8d5a11122c.
Besides this patch contains a better implementation of the generic
handler function handler::multi_range_read_info_const() that
takes into account gaps between ranges when calculating the cost of
range index scans. It also contains some corrections of the
implementation of the handler function records_in_range() for MyISAM.
This patch supports the feature for InnoDB and MyISAM.
MDEV-9931 introduced a counter for keeping track of reads of the
first page of InnoDB data files, because the original implementation
of data-at-rest-encryption for InnoDB introduced new code paths for
reading the pages.
Ultimately, the extra reads of the first page were removed, and
the encryption subsystem will be initialized whenever we first read
the first page of each data file, in fil_node_open_file(). It should not
be that interesting to observe how many times an InnoDB data file was
opened for the first time.
There is only one redo log subsystem in InnoDB. Allocate the object
statically, to avoid unnecessary dereferencing of the pointer.
log_t::create(): Renamed from log_sys_init().
log_t::close(): Renamed from log_shutdown().
log_t::checkpoint_buf_ptr: Remove. Allocate log_t::checkpoint_buf
statically.
There is only one lock_sys. Allocate it statically in order to avoid
dereferencing a pointer whenever accessing it. Also, align some
members to their own cache line in order to avoid false sharing.
lock_sys_t::create(): The deferred constructor.
lock_sys_t::close(): The early destructor.
trx_sys_t::rseg_history_len: Make private, and clarify the
documentation.
trx_sys_t::history_size(): Read rseg_history_len.
trx_sys_t::history_insert(), trx_sys_t::history_remove(),
trx_sys_t::history_add(): Update rseg_history_len.
There is only one transaction system object in InnoDB.
Allocate the storage for it at link time, not at runtime.
lock_rec_fetch_page(): Use the correct fetch mode BUF_GET.
Pages may never be deallocated from a tablespace while
record locks are pointing to them.
Define UNIV_WORD_SIZE as a simple alias to SIZEOF_SIZE_T.
In MariaDB 10.0 and 10.1, it was incorrectly defined as 4 on
64-bit Windows.
MONITOR_OS_PENDING_READS, MONITOR_OS_PENDING_WRITES: Enable by default.
os_n_pending_reads, os_n_pending_writes: Remove.
Use the monitor counters instead.
Allow 64-bit atomic operations on 32-bit systems,
only relying on HAVE_ATOMIC_BUILTINS_64, disregarding
the width of the register file.
Define UNIV_WORD_SIZE correctly on all systems, including Windows.
In MariaDB 10.0 and 10.1, it was incorrectly defined as 4 on
64-bit Windows.
Define HAVE_ATOMIC_BUILTINS_64 on Windows
(64-bit atomics are available on both 32-bit and 64-bit Windows
platforms; the operations were unnecessarily disabled even on
64-bit Windows).
MONITOR_OS_PENDING_READS, MONITOR_OS_PENDING_WRITES: Enable by default.
os_file_n_pending_preads, os_file_n_pending_pwrites,
os_n_pending_reads, os_n_pending_writes: Remove.
Use the monitor counters instead.
os_file_count_mutex: Remove. On a system that does not support
64-bit atomics, monitor_mutex will be used instead.
Also, remove empty .ic files that were not removed by my MySQL commit.
Problem:
InnoDB used to support a compilation mode that allowed to choose
whether the function definitions in .ic files are to be inlined or not.
This stopped making sense when InnoDB moved to C++ in MySQL 5.6
(and ha_innodb.cc started to #include .ic files), and more so in
MySQL 5.7 when inline methods and functions were introduced
in .h files.
Solution:
Remove all references to UNIV_NONINL and UNIV_MUST_NOT_INLINE from
all files, assuming that the symbols are never defined.
Remove the files fut0fut.cc and ut0byte.cc which only mattered when
UNIV_NONINL was defined.
The InnoDB adaptive hash index is sometimes degrading the performance of
InnoDB, and it is sometimes disabled to get more consistent performance.
We should have a compile-time option to disable the adaptive hash index.
Let us introduce two options:
OPTION(WITH_INNODB_AHI "Include innodb_adaptive_hash_index" ON)
OPTION(WITH_INNODB_ROOT_GUESS "Cache index root block descriptors" ON)
where WITH_INNODB_AHI always implies WITH_INNODB_ROOT_GUESS.
As part of this change, the misleadingly named function
trx_search_latch_release_if_reserved(trx) will be replaced with the macro
trx_assert_no_search_latch(trx) that will be empty unless
BTR_CUR_HASH_ADAPT is defined (cmake -DWITH_INNODB_AHI=ON).
We will also remove the unused column
INFORMATION_SCHEMA.INNODB_TRX.TRX_ADAPTIVE_HASH_TIMEOUT.
In MariaDB Server 10.1, it used to reflect the value of
trx_t::search_latch_timeout which could be adjusted during
row_search_for_mysql(). In 10.2, there is no such field.
Other than the removal of the unused column TRX_ADAPTIVE_HASH_TIMEOUT,
this is an almost non-functional change to the server when using the
default build options.
Some tests are adjusted so that they will work with both
-DWITH_INNODB_AHI=ON and -DWITH_INNODB_AHI=OFF. The test
innodb.innodb_monitor has been renamed to innodb.monitor
in order to track MySQL 5.7, and the duplicate tests
sys_vars.innodb_monitor_* are removed.
Problem was that implementation merged from 10.1 was incompatible
with InnoDB 5.7.
buf0buf.cc: Add functions to return should we punch hole and
how big.
buf0flu.cc: Add written page to IORequest
fil0fil.cc: Remove unneeded status call and add test is
sparse files and punch hole supported by file system when
tablespace is created. Add call to get file system
block size. Used file node is added to IORequest. Added
functions to check is punch hole supported and setting
punch hole.
ha_innodb.cc: Remove unneeded status variables (trim512-32768)
and trim_op_saved. Deprecate innodb_use_trim and
set it ON by default. Add function to set innodb-use-trim
dynamically.
dberr.h: Add error code DB_IO_NO_PUNCH_HOLE
if punch hole operation fails.
fil0fil.h: Add punch_hole variable to fil_space_t and
block size to fil_node_t.
os0api.h: Header to helper functions on buf0buf.cc and
fil0fil.cc for os0file.h
os0file.h: Remove unneeded m_block_size from IORequest
and add bpage to IORequest to know actual size of
the block and m_fil_node to know tablespace file
system block size and does it support punch hole.
os0file.cc: Add function punch_hole() to IORequest
to do punch_hole operation,
get the file system block size and determine
does file system support sparse files (for punch hole).
page0size.h: remove implicit copy disable and
use this implicit copy to implement copy_from()
function.
buf0dblwr.cc, buf0flu.cc, buf0rea.cc, fil0fil.cc, fil0fil.h,
os0file.h, os0file.cc, log0log.cc, log0recv.cc:
Remove unneeded write_size parameter from fil_io
calls.
srv0mon.h, srv0srv.h, srv0mon.cc: Remove unneeded
trim512-trim32678 status variables. Removed
these from monitor tests.
The InnoDB source code contains quite a few references to a closed-source
hot backup tool which was originally called InnoDB Hot Backup (ibbackup)
and later incorporated in MySQL Enterprise Backup.
The open source backup tool XtraBackup uses the full database for recovery.
So, the references to UNIV_HOTBACKUP are only cluttering the source code.
Analysis: By design InnoDB was reading first page of every .ibd file
at startup to find out is tablespace encrypted or not. This is
because tablespace could have been encrypted always, not
encrypted newer or encrypted based on configuration and this
information can be find realible only from first page of .ibd file.
Fix: Do not read first page of every .ibd file at startup. Instead
whenever tablespace is first time accedded we will read the first
page to find necessary information about tablespace encryption
status.
TODO: Add support for SYS_TABLEOPTIONS where all table options
encryption information included will be stored.
Contains also:
MDEV-10549 mysqld: sql/handler.cc:2692: int handler::ha_index_first(uchar*): Assertion `table_share->tmp_table != NO_TMP_TABLE || m_lock_type != 2' failed. (branch bb-10.2-jan)
Unlike MySQL, InnoDB still uses THR_LOCK in MariaDB
MDEV-10548 Some of the debug sync waits do not work with InnoDB 5.7 (branch bb-10.2-jan)
enable tests that were fixed in MDEV-10549
MDEV-10548 Some of the debug sync waits do not work with InnoDB 5.7 (branch bb-10.2-jan)
fix main.innodb_mysql_sync - re-enable online alter for partitioned innodb tables
Contains also
MDEV-10547: Test multi_update_innodb fails with InnoDB 5.7
The failure happened because 5.7 has changed the signature of
the bool handler::primary_key_is_clustered() const
virtual function ("const" was added). InnoDB was using the old
signature which caused the function not to be used.
MDEV-10550: Parallel replication lock waits/deadlock handling does not work with InnoDB 5.7
Fixed mutexing problem on lock_trx_handle_wait. Note that
rpl_parallel and rpl_optimistic_parallel tests still
fail.
MDEV-10156 : Group commit tests fail on 10.2 InnoDB (branch bb-10.2-jan)
Reason: incorrect merge
MDEV-10550: Parallel replication can't sync with master in InnoDB 5.7 (branch bb-10.2-jan)
Reason: incorrect merge
