The code in Type_handler_blob****::make_conversion_table_field()
erroneously assumed that row format replication uses
MYSQL_TYPE_TINYBLOB, MYSQL_TYPE_BLOB, MYSQL_TYPE_MEDIUMBLOB,
MYSQL_TYPE_LONGBLOB type codes to tranfer BLOB variations.
In fact, all BLOB variations use MYSQL_TYPE_BLOB as the type
code, while the BLOB packlength (1,2,3 or 4) it tranferred
in metadata.
The bug was introduced by aee068085d
(MDEV-9238 Wrap create_virtual_tmp_table() into a class, split into different steps)
If innodb_fast_shutdown<2, all transactions of active connections
will be rolled back on shutdown. This can take a long time, and
the systemd shutdown timeout should be extended during the wait.
logs_empty_and_mark_files_at_shutdown(): Extend the timeout when
waiting for other threads to complete.
This reverts commit 76ec37f522.
This behaviour change will be done separately in:
MDEV-15832 With innodb_fast_shutdown=3, skip the rollback
of connected transactions
Table_open_cache gets adjusted on server startup to prevent an out of
file descriptor error. However this means that when we reset its value
to default, it does not get re-adjusted. This leads to the mtr
consistency check to fail with different server status at the end of the
test case as opposed to when it started.
To fix the problem, do not make use of the DEFAULT keyword to set the
variable back, instead save the value before any modifications and
restore it from there.
srv_purge_should_exit(): Remove the parameter n_purged.
If we happened to have n_purged==0 while some transaction was still
active, and then that transaction was added to the history list,
we were prematurely stopping the purge. It is more appropriate to
first check for trx_sys.any_active_transactions() == 0
(this count can only decrease during shutdown) and then for
trx_sys.history_size() == 0 (that count typically decreases, but
can increase when any remaining active transactions are committed
or rolled back).
innodb.dml_purge: Remove a server restart, and explicitly wait for
purge, and use FLUSH TABLE FOR EXPORT to read the file contents.
This will make the test run faster, easier to debug, and also
allow it to run with --embedded. This might also help repeat
MDEV-11802 better. The issue MDEV-13603 remains will remain tested
by innodb.table_flags.
In async IO completion code, after reading a page,Innodb can wait for
completion of other bufferpool reads.
This is for example what happens if change-buffering is active.
Innodb on Windows could deadlock, as it did not have dedicated threads
for processing change buffer asynchronous reads.
The fix for that is to have windows now has the same background threads,
including dedicated thread for ibuf, and log AIOs.
The ibuf/read completions are now dispatched to their threads with
PostQueuedCompletionStatus(), the write and log completions are processed
in thread where they arrive.
purge_sys_t::n_submitted: Document that it is only accessed by
srv_purge_coordinator_thread.
purge_sys_t::n_completed: Exclusively use my_atomic access.
srv_task_execute(): Simplify the code.
srv_purge_coordinator_thread(): Test the cheaper condition first.
trx_purge(): Atomically access purge_sys.n_completed.
Remove some code duplication.
trx_purge_wait_for_workers_to_complete(): Atomically access
purge_sys.n_completed. Remove an unnecessary local variable.
trx_purge_stop(): Remove a redundant assignment.
row_ins_sec_index_entry(): Compare a pointer to fil_system.sys_space,
not to a numeric constant. This code was recently changed in MDEV-13637,
and the condition was essentially disabled, potentially causing the
change buffer to grow uncontrollably when something is inserted into
a table that has secondary indexes and resides in the system tablespace.
Thanks to Daniel Black for pointing out that clang 7 flagged a warning
for the comparison of a pointer to an integer.
row_import_for_mysql(): Fix a possible compiler warning.
fil_crypt_read_crypt_data(): Do not attempt to read the tablespace
if the file is unaccessible due to a pending DDL operation, such as
renaming the file or DROP TABLE or TRUNCATE TABLE. This is only
reducing the probability of the race condition, not completely
preventing it.
There is no need for MariaDB Backup to be versioned. Other packages are
very unlikely to need to depend on a certain generation of this package.
Also for MariaDB Backup to run, it does not need the server, but client,
and from client only core parts.
Adjust the fix for MDEV-15472:
In close_cached_tables(), locked_tables_list.reopen_tables(...) call
might be made when a previous call to some function has already returned
error.
In this scenario, the function should return 0, even if reopen_tables(...)
call has succeeded.
When "FLUSH TABLE ... FOR EXPORT" fails, the SQL layer should rollback
the statement. Otherwise we hit an assert when we try to close the
tables while having a non-empty list of statement transaction participants.
dict_sys_tables_type_valid(): Do not reject NO_ROLLBACK (sequence)
in ROW_FORMAT=REDUNDANT.
dict_load_table_low(): When flagging an error, assign *table = NULL.
Failure to do so could cause a crash if an error was flagged when
accessing INFORMATION_SCHEMA.INNODB_SYS_TABLES.
The test tends to fail if many parallel instances of it are executed:
```
mysqltest: At line 23: query 'ALTER TABLE t1 ADD PRIMARY KEY (f1)' failed:
1317: Query execution was interrupted
```
The `ALTER` fails because it is BF aborted due to an earlier `INSERT SELECT`
that is being applied:
```
INSERT INTO t1 (f1) SELECT ...
--connection node_2
SET GLOBAL wsrep_desync = TRUE;
SET SESSION wsrep_on = FALSE;
ALTER TABLE t1 ADD PRIMARY KEY (f1);
SET SESSION wsrep_on = TRUE;
SET GLOBAL wsrep_desync = FALSE;
```
And because the `ALTER` is executed with `wsrep_on = OFF`, it does not
run in total order isolation.
To avoid the problem it must be ensured that the `ALTER` only after the
large `INSERT SELECT` is done. To do so it is sufficient to issue
`SELECT COUNT(*) FROM t1;` from `node_2` before turning off wsrep.
The `SELECT` will trigger `wsrep_sync_wait` and proceed only after the
`INSERT SELECT` from node_1 is done.
Problem was that key rotation from encrypted to unecrypted was skipped
when encryption is disabled (i.e. set global innodb-encrypt-tables=OFF).
fil_crypt_needs_rotation
If encryption is disabled (i.e. innodb-encrypt-tables=off)
and there is tablespaces using default encryption (e.g.
system tablespace) that are still encrypted state we need
to rotate them from encrypted state to unencrypted state.
Boost includes sys/param.h on FreeBSD, which in turn defines setbit()
macro. This macro is conflicting with open_query::judy_bitset::setbit().
Reordered includes such that oqgraph_judy.h never sees this macro.
Also removed duplicate includes of graphcore-config.h, which is included
by graphcore-graph.h/oqgraph_shim.h/oqgraph_thunk.h.
buf_flush_remove(): Disable the output for now, because we
certainly do not want this after every page flush on shutdown.
It must be rate-limited somehow. There already is a timeout
extension for waiting the page cleaner to exit in
logs_empty_and_mark_files_at_shutdown().
log_write_up_to(): Use correct format.
srv_purge_should_exit(): Move the timeout extension to the
appropriate place, from one of the callers.
