Do not execute user callback just after pwrite. Instead, submit user
function as task into thread pool. This way, the IO thread would not hog
aiocb, which is a limited (in Innodb) resource
Add fixed for tests mysqld--help,aix.rdiff and sysvars_server_notembedded,aix.rdiff
AIX couldn't compile in embedded mode so leaving sysvars_server_embedded
for later (if required).
Additional improvements aimed at improving operational
reliability of the SST scripts:
1) Script need to give rsync and stunnel a short time to
terminate after "kill -9" before the first PID check
using ps utility;
2) The temporary file used to create the binlog index could
sometimes remain in the data directory if tar failed and
then may be reused without being cleaned up (the next
time when SST was run) - now it's fixed;
3) The temporary file used to build the binlog index is now
created using mktemp and, if this variable is present in
the configuration file, in tmpdir;
4) Checking the secret tag in SST via rsync is made faster
and does not require creating a temporary file, which
could remain in the data directory in case of failure;
5) Added "-F" option to grep to check the tag when using
mariabackup/xtrabackup-v2 - to avoid possible collisions
in case of special characters in the tag value (unlikely
scenario, but the new check is more reliable).
This commit fixes a bug was originally discovered during the
galera_nbo_sst_slave mtr test for 10.6 branch. However it is
relevant for all versions and can lead to intermittent SST
crashes via rsync on very fast server restarts - when a new
SST process (for example, after starting a new server instance)
overlaps the old SST process started by the previous, already
terminated server. This overlap can result in the new rsync
being killed instead of the old rsync, or the pid file from
the new rsync being killed, which then lead to problems.
- During online alter conversion from compact to redundant,
virtual column field length already set during
innobase_get_computed_value(). Skip the char(n) check for
virtual column in row_merge_buf_add()
- InnoDB fails to check DB_COMPUTE_VALUE_FAILED error in
row_merge_read_clustered_index() and wrongly asserts that
the buffer shouldn't be ran out of memory. Alter table
should give warning when the column value is being
truncated.
Problem:
=======
- InnoDB iterates the fil_system space list to encrypt the
tablespace in case of key rotation. But it is not
necessary for any encryption plugin which doesn't do
key version rotation.
Solution:
=========
- Introduce a new variable called srv_encrypt_rotate to
indicate whether encryption plugin does key rotation
fil_space_crypt_t::key_get_latest_version(): Enable the
srv_encrypt_rotate only once if current key version is
higher than innodb_encyrption_rotate_key_age
fil_crypt_must_default_encrypt(): Default encryption tables
should be added to default_encryp_tables list if
innodb_encyrption_rotate_key_age is zero and encryption
plugin doesn't do key version rotation
fil_space_create(): Add the newly created space to
default_encrypt_tables list if
fil_crypt_must_default_encrypt() returns true
Removed the nondeterministic select from
innodb-key-rotation-disable test. By default,
InnoDB adds the tablespace to the rotation list and
background crypt thread does encryption of tablespace.
So these select doesn't give reliable results.
- Rename the rotation list to default_encrypt_tables in
fil_system_t. Because rotation list naming could be
misleading when it comes to key version rotation
- Rename is_in_rotation_list to is_in_default_encrypt
in fil_space_t
- Rename keyrotate_next function to default_encrypt_next
fil_system_t::default_encrypt_next(): Find the next
suitable default encrypt table if beginning of
default_encrypt_tables list has been scheduled
to be deleted
Brad Smith made this OpenBSD file based of the
FreeBSD cmake directives in commit ab58904367
by the Monty Program Ab.
As such the Oracle Copyright header isn't really applicable.
* read_command_buf is a pointer now, sizeof() no longer reflects its
length, read_command_buflen is.
* my_safe_print_str() prints multiple screens of '\0' bytes after the
query end and up to read_command_buflen. Use fprintf() instead.
* when setting connection->name to "-closed_connection-" update
connection->name_len to match.
The check-testcase record uses a mysqltest connection
to the database to do the recording. With the server configured
as an abstract socket, the mysqltest client cannot connect and
fails.
We work around this by starting the server as normal and then
restart with an abstract socket and test this.
This didn't affect Windows as it just did a tcp connection.
So this did affect all unix socket based systems except Linux
as this was the only one that supported abstract sockets.
Not all environments have 'diff' installed. Most notably CentOS 8
does not have diff out-of-the-box. Thus users running 'cmake .' and
'make' would fail to build MariaDB, and they would think the error
was in ABI incompatibilities due to the error message emitted by CMake
when in reality simply 'diff' was missing.
This fixes it and makes the developer experience better by simply
skipping the diffing if 'diff' is not found.
Closes#1846
Removed Field_map, since it was used only in a single function.
Fixed is_indexed_agg_distinct(), since it relied on initialization of
Bitmap in constructor.
Fixes MDEV-25888 in 10.4
the idea of main.failed_auth_unixsocket was to have existing
user account (root) authenticate with unix_socket, then login with
non-existent user name, Non-existent user name forces the server
to perform the authentication in the name of some random existing
user. But it must still fail at the end, as the user name is wrong.
In 10.4 a second predefined user was added, mariadb.sys, so root
is not the only user in mysql.global_priv and unix_socket auth
must be forced for all existing user accounts, because we cannot
know what user account the server will randomly pick for non-existing
user auth.
RocksDB failed to package on fedora 32 and 33 with
*** ERROR: ambiguous python shebang in /usr/bin/myrocks_hotbackup: #!/usr/bin/env python. Change it to python3 (or python2) explicitly.
When doing a ALTER TABLE ... RENAME, MariaDB doesn't rename
original table to #sql-backup, which it does in other cases,
but insteads drops the original table directly. However
this optimization doesn't work in case of InnoDB table
with a foreign key constraint.
During copy algorithm, InnoDB fails to rename the foreign key
constraint(MDEV-25855). With this optimisation, InnoDB also
fails to drop the original table because the table has
FOREIGN Key constraint exist in INNODB_SYS_FOREIGN table.
This leads to orphan .ibd file in InnoDB dictionary.
so disabling this optimization when FK is involved.
Reviewer: monty@mariadb.org
* Make Item_in_optimizer::fix_fields inherit the with_window_func
attribute of the subquery's left expression (the subquery itself
cannot have window functions that are aggregated in this select)
* Make Item_cache_wrapper::Item_cache_wrapper() inherit
with_window_func attribute of the item it is caching.
Occasionally, the test innodb.alter_copy would fail in MariaDB 10.6.1,
reporting DB_MISSING_HISTORY during CHECK TABLE. It started to occur during
the development of MDEV-25180, which introduced purge_sys.stop_SYS().
If we delay purge more during DDL operations, then the test would
almost always fail. The reason is that during startup we will restore
a purge view, and CHECK TABLE would still use REPEATABLE READ
even though innodb_read_only is set and other isolation levels
than READ UNCOMMITTED are not guaranteed to work.
ha_innobase::check(): Use READ UNCOMMITTED isolation level if
innodb_read_only is set or innodb_force_recovery exceeds 3.
dict_set_corrupted(): Do not update the persistent data dictionary
if innodb_force_recovery exceeds 3.
buf_read_ibuf_merge_pages(): If space->size is 0, invoke
fil_space_get_size() to determine the size of the tablespace
by reading the header page. Only after that proceed to delete
any entries that are beyond the end of the tablespace.
Otherwise, we could be deleting valid entries that actually
need to be applied.
This fixes a regression that had been introduced in
commit b80df9eba2 (MDEV-21069),
which aimed to avoid crashes during DROP TABLE of corrupted tables.
If a select query contained an ORDER BY clause that followed a LIMIT clause
or an ORDER BY clause or ORDER BY with LIMIT the EXPLAIN output for the
query showed an execution plan different from that was actually executed.
Approved by Roman Nozdrin <roman.nozdrin@mariadb.com>
This happended when an aria table was already used by the system before
running zerofill, which could happen with Aria system tables.
Fixed by using a proper page type when reading pages in zerofill
ibuf_read_merge_pages(): Disable some code that was added in MDEV-20394
in order to avoid a server hang if the change buffer is corrupted,
presumably due to the race condition in recovery that was later fixed in
MDEV-24449. The code will still be available in debug builds when
the command line option --debug=d,ibuf_merge_corruption is specified.
Due to MDEV-19514, the impact of this code is much worse starting
with the 10.5 series. In older versions, the code was only enabled
during a shutdown with innodb_fast_shutdown=0, but in 10.5 it was
active during the normal operation of the server.
This is because on different compilation and server configurations the
memory usage is different and the query can get killed in different places
with different error messages as a result.
Reviewer: None (trival change)
This partially reverts commit d7321893d8.
The *.jar files are not being built and all Debian builds are failing
as dh_install stops on missing files. To build them we would need to also
add new Java build dependencies.
In a stable release (10.2->10.5) we shouldn't add new files and certainly
not any new build dependencies, so reverting commit.
Also, the files are located in a different path, and already included
in the mariadb-test-data package:
/usr/share/mysql/mysql-test/plugin/connect/connect/std_data/JavaWrappers.jar
/usr/share/mysql/mysql-test/plugin/connect/connect/std_data/JdbcMariaDB.jar
/usr/share/mysql/mysql-test/plugin/connect/connect/std_data/Mongo2.jar
/usr/share/mysql/mysql-test/plugin/connect/connect/std_data/Mongo3.jar
This change needs to be redesigned and applies only on 10.6 or newer.
* Clean up autobake-deb.sh
- No need to define any TokuDB rules, there is no such package
- No need to define RocksDB arch, it already has "Architecture:" line
- No need to dh-systemd backwards compat stanza, neither Debian Jessie
nor Ubuntu Xenial has any new MariaDB 10.5 releases anymore
- Minor spelling fixes
* Ensure dch runs non-interactively so builds pass with new dch version
A recent version of dch (available in Ubuntu Hirsute and Debian Bullseye)
had a change in behaviour that it started prompting if the DEBEMAIL or
EMAIL variable as unset, asking for confirmation. We can't have anything
interactive in our build scripts, so prevent this prompt by giving
--controlmaint to the command, so it always uses the name and email from
the debian/control file and does not prompt anything.
The command-line argument has been around for a long time, so it is safe
to use on all Debian/Ubuntu builds we have.
See https://manpages.debian.org/jessie/devscripts/dch.1.en.html
Since MariaDB 10.5 is the oldest release we still release for Ubuntu Hisute
and Debian Bullseye, merge this on 10.5 and from there merge up to latest.
No need to consider 10.2, 10.3 and 10.4 as those will not be released for
Ubuntu Bullseye or Ubuntu Hirsute.
* Minor Salsa-CI cleanup
- Fix spelling (synced from downstream Debian)
* Many minor spelling fixes (synced from downstream Debian)
ha_heap::external_lock contains some consistency checks for the table,#
in a debug compilation.
This code suffers from lack of synchronization, in a rare case
where mysql_lock_tables() fail, and unlock is forced, even if lock was
not previously taken.
To workaround, require EXTRA_DEBUG compile definition in order to activate
the consistency checks.The code still might be useful in some cases - but
the audience are developers looking for errors in single-threaded scenarios,
rather than multiuser stress-tests.
