The error handling in the MDEV-13564 TRUNCATE TABLE was broken
when an error occurred during table creation.
row_create_index_for_mysql(): Do not drop the table on error.
fts_create_one_common_table(), fts_create_one_index_table():
Do drop the table on error.
create_index(), create_table_info_t::create_table():
Let the caller handle the index creation errors.
ha_innobase::create(): If create_table_info_t::create_table()
fails, drop the incomplete table, roll back the transaction,
and finally return an error to the caller.
The relevant InnoDB/XtraDB fixes up to 5.6.42 had already
been applied to MariaDB in commit 30c3d6db32.
Revert some changes that appeared in
the merge commit 87d852f102.
- Backported the MYSQL_SYSVAR_SIZE_T to 10.0
- The parameter innodb_ft_result_cache_limit was only 32 bits wide
also on 64-bit systems. Make it size_t, so that it will be 64 bits
on 64-bit systems.
- Added a test case that show how innodb_ft_result_cache_limit variables
behaves in 32bit and 64 bit system.
When converting table identifiers to a new format,
some tables can be renamed twice, which subsequently
leads to the appearance of "false" auxiliary tables
belonging to another main (parent) table (which does
not actually have auxiliary tables).
This is because the table number is repeatedly added
to the aux_tables_to_rename vector inside the function
fts_check_and_drop_orphaned_tables.
To correct this error, we must add a check for the
occurrence of the table number in the aux_tables_to_rename
vector before adding a new element.
https://jira.mariadb.org/browse/MDEV-16656
Implement undo tablespace truncation via normal redo logging.
Implement TRUNCATE TABLE as a combination of RENAME to #sql-ib name,
CREATE, and DROP.
Note: Orphan #sql-ib*.ibd may be left behind if MariaDB Server 10.2
is killed before the DROP operation is committed. If MariaDB Server 10.2
is killed during TRUNCATE, it is also possible that the old table
was renamed to #sql-ib*.ibd but the data dictionary will refer to the
table using the original name.
In MariaDB Server 10.3, RENAME inside InnoDB is transactional,
and #sql-* tables will be dropped on startup. So, this new TRUNCATE
will be fully crash-safe in 10.3.
ha_mroonga::wrapper_truncate(): Pass table options to the underlying
storage engine, now that ha_innobase::truncate() will need them.
rpl_slave_state::truncate_state_table(): Before truncating
mysql.gtid_slave_pos, evict any cached table handles from
the table definition cache, so that there will be no stale
references to the old table after truncating.
== TRUNCATE TABLE ==
WL#6501 in MySQL 5.7 introduced separate log files for implementing
atomic and crash-safe TRUNCATE TABLE, instead of using the InnoDB
undo and redo log. Some convoluted logic was added to the InnoDB
crash recovery, and some extra synchronization (including a redo log
checkpoint) was introduced to make this work. This synchronization
has caused performance problems and race conditions, and the extra
log files cannot be copied or applied by external backup programs.
In order to support crash-upgrade from MariaDB 10.2, we will keep
the logic for parsing and applying the extra log files, but we will
no longer generate those files in TRUNCATE TABLE.
A prerequisite for crash-safe TRUNCATE is a crash-safe RENAME TABLE
(with full redo and undo logging and proper rollback). This will
be implemented in MDEV-14717.
ha_innobase::truncate(): Invoke RENAME, create(), delete_table().
Because RENAME cannot be fully rolled back before MariaDB 10.3
due to missing undo logging, add some explicit rename-back in
case the operation fails.
ha_innobase::delete(): Introduce a variant that takes sqlcom as
a parameter. In TRUNCATE TABLE, we do not want to touch any
FOREIGN KEY constraints.
ha_innobase::create(): Add the parameters file_per_table, trx.
In TRUNCATE, the new table must be created in the same transaction
that renames the old table.
create_table_info_t::create_table_info_t(): Add the parameters
file_per_table, trx.
row_drop_table_for_mysql(): Replace a bool parameter with sqlcom.
row_drop_table_after_create_fail(): New function, wrapping
row_drop_table_for_mysql().
dict_truncate_index_tree_in_mem(), fil_truncate_tablespace(),
fil_prepare_for_truncate(), fil_reinit_space_header_for_table(),
row_truncate_table_for_mysql(), TruncateLogger,
row_truncate_prepare(), row_truncate_rollback(),
row_truncate_complete(), row_truncate_fts(),
row_truncate_update_system_tables(),
row_truncate_foreign_key_checks(), row_truncate_sanity_checks():
Remove.
row_upd_check_references_constraints(): Remove a check for
TRUNCATE, now that the table is no longer truncated in place.
The new test innodb.truncate_foreign uses DEBUG_SYNC to cover some
race-condition like scenarios. The test innodb-innodb.truncate does
not use any synchronization.
We add a redo log subformat to indicate backup-friendly format.
MariaDB 10.4 will remove support for the old TRUNCATE logging,
so crash-upgrade from old 10.2 or 10.3 to 10.4 will involve
limitations.
== Undo tablespace truncation ==
MySQL 5.7 implements undo tablespace truncation. It is only
possible when innodb_undo_tablespaces is set to at least 2.
The logging is implemented similar to the WL#6501 TRUNCATE,
that is, using separate log files and a redo log checkpoint.
We can simply implement undo tablespace truncation within
a single mini-transaction that reinitializes the undo log
tablespace file. Unfortunately, due to the redo log format
of some operations, currently, the total redo log written by
undo tablespace truncation will be more than the combined size
of the truncated undo tablespace. It should be acceptable
to have a little more than 1 megabyte of log in a single
mini-transaction. This will be fixed in MDEV-17138 in
MariaDB Server 10.4.
recv_sys_t: Add truncated_undo_spaces[] to remember for which undo
tablespaces a MLOG_FILE_CREATE2 record was seen.
namespace undo: Remove some unnecessary declarations.
fil_space_t::is_being_truncated: Document that this flag now
only applies to undo tablespaces. Remove some references.
fil_space_t::is_stopping(): Do not refer to is_being_truncated.
This check is for tablespaces of tables. Potentially used
tablespaces are never truncated any more.
buf_dblwr_process(): Suppress the out-of-bounds warning
for undo tablespaces.
fil_truncate_log(): Write a MLOG_FILE_CREATE2 with a nonzero
page number (new size of the tablespace in pages) to inform
crash recovery that the undo tablespace size has been reduced.
fil_op_write_log(): Relax assertions, so that MLOG_FILE_CREATE2
can be written for undo tablespaces (without .ibd file suffix)
for a nonzero page number.
os_file_truncate(): Add the parameter allow_shrink=false
so that undo tablespaces can actually be shrunk using this function.
fil_name_parse(): For undo tablespace truncation,
buffer MLOG_FILE_CREATE2 in truncated_undo_spaces[].
recv_read_in_area(): Avoid reading pages for which no redo log
records remain buffered, after recv_addr_trim() removed them.
trx_rseg_header_create(): Add a FIXME comment that we could write
much less redo log.
trx_undo_truncate_tablespace(): Reinitialize the undo tablespace
in a single mini-transaction, which will be flushed to the redo log
before the file size is trimmed.
recv_addr_trim(): Discard any redo logs for pages that were
logged after the new end of a file, before the truncation LSN.
If the rec_list becomes empty, reduce n_addrs. After removing
any affected records, actually truncate the file.
recv_apply_hashed_log_recs(): Invoke recv_addr_trim() right before
applying any log records. The undo tablespace files must be open
at this point.
buf_flush_or_remove_pages(), buf_flush_dirty_pages(),
buf_LRU_flush_or_remove_pages(): Add a parameter for specifying
the number of the first page to flush or remove (default 0).
trx_purge_initiate_truncate(): Remove the log checkpoints, the
extra logging, and some unnecessary crash points. Merge the code
from trx_undo_truncate_tablespace(). First, flush all to-be-discarded
pages (beyond the new end of the file), then trim the space->size
to make the page allocation deterministic. At the only remaining
crash injection point, flush the redo log, so that the recovery
can be tested.
fts_query(): Remove a redundant condition (result will never be NULL),
and instead check if *result is NULL, to prevent SIGSEGV in
fts_query_free_result().
The functions fts_ast_visit() and fts_query() inside
InnoDB FULLTEXT INDEX query processing are not checking
for THD::killed (trx_is_interrupted()), like anything
that potentially takes a long time should do.
This is a port of the following change from MySQL 5.7.23,
with a completely rewritten test case.
commit c58c6f8f66ddd0357ecd0c99646aa6bf1dae49c8
Author: Aakanksha Verma <aakanksha.verma@oracle.com>
Date: Fri May 4 15:53:13 2018 +0530
Bug #27155294 MAX_EXECUTION_TIME NOT INTERUPTED WITH FULLTEXT SEARCH USING MECAB
This is a backport of the following fix from MySQL 5.7.23.
Some code refactoring has been omitted, and the test case has
been adapted to MariaDB.
commit 7a689acaa65e9d602575f7aa53fe36a64a07460f
Author: Krzysztof Kapuścik <krzysztof.kapuscik@oracle.com>
Date: Tue Mar 13 12:34:03 2018 +0100
Bug#27082268 Invalid FTS sync synchronization
The fix closes two issues:
Bug #27082268 - INNODB: FAILING ASSERTION: SYM_NODE->TABLE != NULL DURING FTS SYNC
Bug #27095935 - DEADLOCK BETWEEN FTS_DROP_INDEX AND FTS_OPTIMIZE_SYNC_TABLE
Both issues were related to a FTS cache sync being done during
operations that perfomed DDL actions on internal FTS tables
(ALTER TABLE, TRUNCATE). In some cases the FTS tables and/or
internal cache structures could get removed while still being
used to perform FTS synchronization leading to crashes. In other
the sync operations could not get finishes as it was waiting for
dict lock which was taken by thread waiting for the background
sync to be finished.
The changes done includes:
- Stopping background operations during ALTER TABLE and TRUNCATE.
- Removal of unused code in FTS.
- Cleanup of FTS sync related code to make it more readable and
easier to maintain.
RB#18262
Problem:
As part of bug #24938374 fix, dict_operation_lock was not taken by
fts_optimize_thread while syncing fts cache.
Due to this change, alter query is able to update SYS_TABLE rows
simultaneously. Now when fts_optimizer_thread goes open index table,
It doesn't open index table if the record corresponding to that table is
set to REC_INFO_DELETED_FLAG in SYS_TABLES and hits an assert.
Fix:
If fts sync is already in progress, Alter query would wait for sync to
complete before renaming table.
RB: #19604
Reviewed by : Jimmy.Yang@oracle.com
PROBLEM
-------
Whenever an fts table is created it registers itself in a queue which
is operated by a background thread whose job is to optimize the
fts tables in background. Additionally we place these fts tables in
non-LRU list so that they cannot be evicted from cache. But in the
scenario when a node is brought up which is already having fts
tables ,we first try to load the fts tables in dictionary ,but we skip
the part where it is added in background queue and in non-LRU list because
the background thread is not yet created,so these tables are loaded
but they can be evicted from the cache. Now coming to the deadlock scenario
1. A Server background thread is trying to evict a table from the cache
because the cache is full,so it scans the LRU list for the tables it can
evict.It finds the fts table (because of the reason explained above)
can be evicted and it takes the dict_sys->mutex (this is a system wide mutex)
submits a request to the background thread to remove this table from queue
and waits it to be completed.
2. In the mean time fts_optimize_thread() is processing another job
in the queue and needs dict_sys->mutex for a small amount of time,
but it cannot get it because it is blocked by the first background thread.
So Thread 1 is waiting for its job to be completed by Thread 2,whereas Thread 2
is waiting for dict_sys->mutex held by thread 1 ,causing the deadlock.
FIX
If creating a secondary index fails (typically, ADD UNIQUE INDEX fails
due to duplicate key), it is possible that concurrently running UPDATE
or DELETE will access the index stub and hit the debug assertion.
It does not make any sense to keep updating an uncommitted index whose
creation has failed.
dict_index_t::is_corrupted(): Replaces dict_index_is_corrupted().
Also take online_status into account.
Replace some calls to dict_index_is_clust() with calls to
dict_index_t::is_primary().
fts_sync(): If the dict_table_t::to_be_dropped flag is set,
do not "goto begin_sync".
Also, clean up the way how dict_index_t::index_fts_syncing
is cleared.
It looks like this regression was introduced by merging
Oracle Bug #24938374 MYSQL CRASHED AFTER LONG WAIT ON DICT OPERATION LOCK
WHILE SYNCING FTS INDEX
068f8261d4
from MySQL 5.6.38 into MariaDB 10.0.33, 10.1.29, 10.2.10.
The same hang is present in MySQL 5.7.20.
fts_cmp_set_sync_doc_id(), fts_load_stopword(): Start the transaction
in read-only mode if innodb_read_only is set.
fts_update_sync_doc_id(), fts_commit_table(), fts_sync(),
fts_optimize_table(): Return DB_READ_ONLY if innodb_read_only is set.
fts_doc_fetch_by_doc_id(), fts_table_fetch_doc_ids():
Remove the code to start an internal transaction or to roll back,
because this is a read-only operation.
If CREATE TABLE...FULLTEXT INDEX was initiated right before shutdown,
then the function fts_load_stopword() could commit modifications
after shutdown was initiated, causing an assertion failure in
the function trx_purge_add_update_undo_to_history().
Mark as internal all the read/write transactions that
modify fulltext indexes, so that they will be ignored by
the assertion that guards against transaction commits
after shutdown has been initiated.
fts_optimize_free(): Invoke trx_commit_for_mysql() just in case,
because in fts_optimize_create() we started the transaction as
internal, and fts_free_for_backgruond() would assert that the
flag is clear. Transaction commit would clear the flag.
If a crash occurs during ALTER TABLE…ALGORITHM=COPY, InnoDB would spend
a lot of time rolling back writes to the intermediate copy of the table.
To reduce the amount of busy work done, a work-around was introduced in
commit fd069e2bb3 in MySQL 4.1.8 and 5.0.2,
to commit the transaction after every 10,000 inserted rows.
A proper fix would have been to disable the undo logging altogether and
to simply drop the intermediate copy of the table on subsequent server
startup. This is what happens in MariaDB 10.3 with MDEV-14717,MDEV-14585.
In MariaDB 10.2, the intermediate copy of the table would be left behind
with a name starting with the string #sql.
This is a backport of a bug fix from MySQL 8.0.0 to MariaDB,
contributed by jixianliang <271365745@qq.com>.
Unlike recent MySQL, MariaDB supports ALTER IGNORE. For that operation
InnoDB must for now keep the undo logging enabled, so that the latest
row can be rolled back in case of an error.
In Galera cluster, the LOAD DATA statement will retain the existing
behaviour and commit the transaction after every 10,000 rows if
the parameter wsrep_load_data_splitting=ON is set. The logic to do
so (the wsrep_load_data_split() function and the call
handler::extra(HA_EXTRA_FAKE_START_STMT)) are joint work
by Ji Xianliang and Marko Mäkelä.
The original fix:
Author: Thirunarayanan Balathandayuthapani <thirunarayanan.balathandayuth@oracle.com>
Date: Wed Dec 2 16:09:15 2015 +0530
Bug#17479594 AVOID INTERMEDIATE COMMIT WHILE DOING ALTER TABLE ALGORITHM=COPY
Problem:
During ALTER TABLE, we commit and restart the transaction for every
10,000 rows, so that the rollback after recovery would not take so long.
Fix:
Suppress the undo logging during copy alter operation. If fts_index is
present then insert directly into fts auxiliary table rather
than doing at commit time.
ha_innobase::num_write_row: Remove the variable.
ha_innobase::write_row(): Remove the hack for committing every 10000 rows.
row_lock_table_for_mysql(): Remove the extra 2 parameters.
lock_get_src_table(), lock_is_table_exclusive(): Remove.
Reviewed-by: Marko Mäkelä <marko.makela@oracle.com>
Reviewed-by: Shaohua Wang <shaohua.wang@oracle.com>
Reviewed-by: Jon Olav Hauglid <jon.hauglid@oracle.com>
The ownership of the field query->intersection usually transfers
to query->doc_ids. In some error scenario, it could be possible
that fts_query_free() would be invoked with query->intersection!=NULL.
Let us handle that case, instead of intentionally crashing the server.
MySQL 5.7 added code to push down the LIMIT to fulltext search
in InnoDB:
commit 2cd0ebf97e1b265e2282d7ceb5d8dfb663ffc48f
Author: Thirunarayanan Balathandayuthapani <thirunarayanan.balathandayuth@oracle.com>
Date: Fri May 27 13:49:28 2016 +0530
Bug #22709692 FTS QUERY EXCEEDS RESULT CACHE LIMIT
The code was disabled when MySQL 5.7.9 was merged to MariaDB 10.2.2.
We shall remove the disabled code and unnecessary variables.
fts_get_next_doc_id(): Assign the first and subsequent FTS_DOC_ID
in the same way: by post-incrementing the cached value.
If there is a user-specified FTS_DOC_ID, do not touch the internal
sequence.
fts_create_doc_id(): Remove.
row_mysql_convert_row_to_innobase(): Implement the logic of
fts_create_doc_id(). Reuse a buffer for the hidden FTS_DOC_ID.
row_get_prebuilt_insert_row(): Allocate a buffer for the hidden
FTS_DOC_ID at the end of prebuilt->ins_upd_rec_buff.
This should affect debug builds only. Debug builds will check that
the status bits of ROW_FORMAT!=REDUNDANT records match the is_leaf
parameter.
The only observable change to non-debug should be the addition of
the is_leaf parameter to the function rec_copy_prefix_to_dtuple(),
and the removal of some calls to update the adaptive hash index
(it is only built for the leaf pages).
This change should have been made in MySQL 5.0.3, instead of
introducing the status flags in the ROW_FORMAT=COMPACT record header.
