Home - Waterfall Grid T-Grid Console Builders Recent Builds Buildslaves Changesources - JSON API - About

Console View


Categories: connectors experimental galera main
Legend:   Passed Failed Warnings Failed Again Running Exception Offline No data

connectors experimental galera main
ffcd09536db8...
ffcd09536db89ce4a1619fe0663db831609d7878
Sergei Golubchik
ssl_info_client without a password hash: COM_QUIT alone if not verified

Without a password hash the client cannot prove that it has verified
the server certificate, it can only say so. It sends COM_QUIT and
the version if it has verified the certificate, and COM_QUIT alone
if it has not.

Assisted-by: Claude:claude-opus-5-5
ff498a14c970...
ff498a14c9707a2b8fe506ac4e729ba5ad14e225
Vladislav Vaintroub
MDEV-38741 mtr: do not suppress the reason of a client disconnect

The global suppressions of "Aborted connection", in mariadb-test-run.pl
and mtr_warnings.sql, hid the line that the client disconnect watcher
writes when it aborts a command: "Aborted connection N to db: 'x' user:
'u' host: 'h' (Client disconnected)". Nothing showed when the watcher
aborted a connection that it should have left alone. main.locale found
one, because its server log is in German and the English entry did not
match.

Make the two entries skip a line that ends in "(Client disconnected)".
Other aborted connections are suppressed as before.

The tests that drop a client on purpose accept the reason:
have_disconnect_watcher.inc, which the tests of the watcher include,
and dirty_close, loaddata, sp_notembedded, ssl_timeout, mysqltest,
innodb_bug38231, innodb.xa_recovery, mysql_client_test,
mysql_client_test_nonblock, mysql_client_test_comp and
rpl_semi_sync_shutdown_await_ack.

A replica closes its connection on purpose when it is stopped, in any
phase of its setup, also while the master runs one of its first
queries. rpl_init.inc accepts the reason for every server of the
replication tests, without writing it to the binlog. A binlog dump
thread logs its disconnect with another message, and only with
log_warnings=4.

Assisted-by: Claude:claude-sonnet-5-5
fe7dd8e92510...
fe7dd8e9251014769613880cd51b6e2baba3976d
Khaled Riyad
MDEV-37335: crash on re-parsing an sp_instr_set that does not own its LEX

A DECLARE of several variables with a common DEFAULT expression creates one
sp_instr_set per variable, all sharing a single LEX that only the last of
them owns. On re-parsing after a metadata change, a non-owning instruction
still has its LEX, and both parse_expr() and validate_lex_and_exec_core()
took a non-null LEX to mean a cursor LEX. The cursor-only code then
dereferenced the nullptr returned by get_lex_for_cursor() and the re-parsed
LEX was never adopted.

Branch on whether the LEX is a cursor LEX instead of whether it is non-null.
e730c177f9ab...
e730c177f9ab908f448bebe3477a57c0f9bfaca7
Vladislav Vaintroub
MDEV-38741 mtr: accept "Client disconnected" in more tests

spider/bugfix.mdev_35959 times out a Spider link on purpose, and
mariabackup.nolock_ddl_during_backup_end ends mariabackup in the middle
of a backup, so both abandon a connection with a statement running.

The Connector/C tests of the unit suite also close connections on
purpose, but cannot call mtr.add_suppression() themselves, so
suite.pm adds the suppression before it starts each test.

Assisted-by: Claude:claude-sonnet-5-5
e1692e483785...
e1692e483785966d1cdd29372b0fc9d6d3fc4e68
Sergei Petrunia
MDEV-38274: Unexpectedly large pages_read_count in main.analyze_stmt_prefetch_count

The testcase checked the values of pages_prefetch_read_count and
pages_read_count that InnoDB reported for the query.

The numbers depend on how fast the prefetch thread is in comparison with
the query thread. It can also happen that we request to prefetch a page
but then still have to wait for it to be read, which is counted in
pages_read_count.

Replace the checks in the file with ones that will hold in all
circumstances.
dc2195422f44...
dc2195422f44a95c1b9575d1c5237ea59f1c3bdb
Khaled Riyad
MDEV-37335: crash on re-parsing an sp_instr_set that does not own its LEX

A DECLARE of several variables with a common DEFAULT expression creates one
sp_instr_set per variable, all sharing a single LEX that only the last of
them owns. On re-parsing after a metadata change, a non-owning instruction
still has its LEX, and both parse_expr() and validate_lex_and_exec_core()
took a non-null LEX to mean a cursor LEX. The cursor-only code then
dereferenced the nullptr returned by get_lex_for_cursor() and the re-parsed
LEX was never adopted.

Branch on whether the LEX is a cursor LEX instead of whether it is non-null.
c92c8bb53203...
c92c8bb532033d53bd50ee8ac8ee02f170d1723e
Sergei Golubchik
MDEV-40530 require_secure_transport=CLIENT

An account without a password hash (PAM, gssapi, empty password)
cannot prove that the client has verified the server certificate,
so with require_secure_transport=ON it cannot connect over TCP,
unless it requires a client certificate.

require_secure_transport=CLIENT trusts such clients, as
require_secure_transport did before ssl_info_client. Accounts with
a password hash still need a valid ssl_info_client.

my_getopt: an enum that starts with OFF,ON accepts TRUE and FALSE,
like a boolean does, so --require-secure-transport=TRUE still works.

Assisted-by: Claude:claude-opus-5-5
c430eb9a370b...
c430eb9a370b4897c316623b42acf9b890b3b380
Daniel Black
MDEV-41848: Atomic CoR changed ER_TRUNCATE_ILLEGAL_FK code and broke DBAL

Restore the use of ER_TRUNCATE_ILLEGAL_FK for when TRUNCATE TABLE
hits a foreign key. It was discovered that application frameworks
depend on the error code number and we cannot regress those.

To make this format of ER_TRUNCATE_ILLEGAL_FK match ER_DDL_ILLEGAL_FK
so that correct error message is used. The translations are changed
to ensure the case of the SQL command looks similar in both messages.

Also the sql_table.cc didn't need the include header that defined
fk_truncate_illegal_if_parent.

Fixes: d0b279f18527ba9188d4d9d95d20a4b415bcdb8d / MDEV-25292
Ref: https://github.com/doctrine/dbal/issues/7532

Review: Monty Widenius
b90f74038789...
b90f74038789ade886e06c20902c754589354091
Oleksandr Byelkin
Merge branch '11.8' into 12.3
b8d7f1594ef5...
b8d7f1594ef58f76bbe90b030fb4ccf90cbc0d63
Monty
Enabling ha_videx causes 'drop table xxx' to not generate errors

Fixed by having videx_hton->drop_table return -1 (no table data attached
to this object)
b880092b1582...
b880092b15824a5de7c74ea90dc5cb620b0cfea6
Vladislav Vaintroub
MDEV-38741 Abort the running statement when its client disconnects

The server notices that a client is gone only when it reads from or
writes to the socket. A statement that waits, for example for an InnoDB
row lock, keeps running, and its transaction keeps its locks.

Watch the client socket from the end of login until the connection ends
or the client sends COM_QUIT, with epoll (Linux), kqueue (BSD, macOS),
or on Windows with WSAEventSelect and an I/O completion port, and the
exit of the client process for a named pipe. When the client has gone,
the watcher sets DISCONNECT_SIGNALED in THD::disconnect_flags. If a
command that expects an answer is running, it is aborted with
KILL_CONNECTION. If the connection is idle, the next such command is
aborted when it starts. COM_QUIT and COM_STMT_CLOSE are not aborted, a
client can send them and then quit.

A command that is only sending its reply is not aborted, there is
nothing left to interrupt: it sets DISCONNECT_UNINTERRUPTIBLE before
the reply is written, and the lost connection is then logged as usual.
The watcher and the connection thread order their accesses with atomic
read-modify-write operations on the flags, no lock is taken for a
command.

The socket is not shut down, the connection ends after the error is
sent, and the reason, "Client disconnected", is logged once. A
connection with the disconnect noted, idle or only finishing its reply,
shows that State in the processlist. Galera connections are not
aborted.

The global variable disconnect_watcher, ON by default, can be set OFF,
which also ends the watches that exist. It exists only where there is a
watcher: Linux, Windows, the BSDs and macOS, not the embedded server.

New tests: main.disconnect_*, sys_vars.disconnect_watcher_basic and
mysql_client_test test_disconnect_*. Some rpl and innodb tests are
adjusted to pass with the watcher ON and OFF.

Assisted-by: Claude:claude-sonnet-5-5
b2836db58913...
b2836db589137de5332e9f387407c13b94bc980c
Oleksandr Byelkin
fix
b0dd0a085ab6...
b0dd0a085ab6b378aecb6a3b0e9d687b5988134e
Vladislav Vaintroub
MDEV-38741 Do not abort COM_STMT_PREPARE after its reply

COM_STMT_PREPARE writes its reply in send_prep_stmt(), not in
Protocol::end_statement(), so the command was not marked uninterruptible
until the reply was on the wire. A client that disconnects right after
the last PREPARE reply, as mysql_client_test does with its prepared
statements, could be seen by the watcher in that window, and a graceful
client was aborted. Mark it before the reply is written.

In mysql_client_test, the pipelined tests accept CR_SERVER_LOST on every
platform: the server closes the connection with the INSERT unread, and
the reset can drop the error. Remove test_disconnect_stmt_close_quit, it
cannot make the watcher see the disconnect at a chosen moment of the
server thread. SHUT_WR is defined for Windows in my_net.h, next to
SHUT_RDWR.

Assisted-by: Claude:claude-sonnet-5-5
ab976e5b770e...
ab976e5b770eb81e7493ec907b61ded83ee9ab53
Sergei Golubchik
skip columnstore better

followup for 550444c3bd4
a793825c9946...
a793825c99460ac1c550e00c68939c3a8161dd10
Oleksandr Byelkin
Merge branch '11.8' into 12.3
8e2919178136...
8e2919178136de95fb0b45dc979ff4426aa9221f
Monty
MDEV-41811 Atomic CoR: a failed CREATE OR REPLACE ... SELECT NEXTVAL() loses the old table

The reason for this failure is commit
bb4dd70e7c6509ed9bf8c38d44dffb6d2fcca415 (MDEV-13005) which generated
a new error code in ha_discover_table() that ddl_log_revert() for
create or replace table cannot handle. The code also did not take
into account that share and thd->lex->query_tables are not guaranteed
to refer to the same table.

I removed the code for handling sequences from ha_discover_table() and
instead generate the correct error in open_table_error(). This
handles many cases the old code did not handle.

Changes:
- share->table_type == TABLE_TYPE_SEQUENCE is set early in open for
  sequences to be able to generate correct error messages
- share->table_type == TABLE_TYPE_VIEW is set for views during open
  This is mainly to make the table_type have an usable value for views.
8a7a28ffc4b9...
8a7a28ffc4b93d02d83e528acca9c80b5469c765
Sergei Golubchik
send ssl_info_client to prove the authentication was not relayed

A MitM can relay the authentication, let the client fail on ssl_info
and keep using the authenticated connection.

* MARIADB_CLIENT_SSL_INFO_CLIENT: set it whenever TLS is used
* after the authentication OK packet send ssl_info_client, the server
  does not reply (but might disconnect in case of active MitM).
80bf819b4e9d...
80bf819b4e9d26d70cb81ee9f05a6c92927726cd
Monty
MDEV-41811 Atomic CoR: a failed CREATE OR REPLACE ... SELECT NEXTVAL() loses the old table

The reason for this failure is commit
bb4dd70e7c6509ed9bf8c38d44dffb6d2fcca415 (MDEV-13005) which generated
a new error code in ha_discover_table() that ddl_log_revert() for
create or replace table cannot handle. The code also did not take
into account that share and thd->lex->query_tables are not guaranteed
to refer to the same table.

I removed the code for handling sequences from ha_discover_table() and
instead generate the correct error in open_table_error(). This
handles many cases the old code did not handle.

Changes:
- share->table_type == TABLE_TYPE_SEQUENCE is set early in open for
  sequences to be able to generate correct error messages
- share->table_type == TABLE_TYPE_VIEW is set for views during open
  This is mainly to make the table_type have an usable value for views.
6a0c09235c17...
6a0c09235c17ccd972de097416e0aa0d66768683
Monty
MDEV-41810 Atomic CoR: a crash at the end of CREATE OR REPLACE loses both the old and the new table

The given test case was already in the test suite, but unfortunately I had
missed a failing test in the result file.

Fixed by marking the created table complete before removing the
backup table.
I also extended create_replace_no_binlog.test with a new crash point to
ensure state is tested before and after marking table complete.
6585231c8b5b...
6585231c8b5b0e4c903aa18099e7d592147cf1b1
Vladislav Vaintroub
MDEV-38741 Abort the running statement when its client disconnects

The server notices that a client is gone only when it reads from or
writes to the socket. A statement that waits, for example for an InnoDB
row lock, keeps running, and its transaction keeps its locks.

Watch the client socket from the end of login until the connection ends
or the client sends COM_QUIT, with epoll (Linux), kqueue (BSD, macOS),
or on Windows with WSAEventSelect and an I/O completion port, and the
exit of the client process for a named pipe. When the client has gone,
the watcher sets DISCONNECT_SIGNALED in THD::disconnect_flags. If a
command that expects an answer is running, it is aborted with
KILL_CONNECTION. If the connection is idle, the next such command is
aborted when it starts. COM_QUIT and COM_STMT_CLOSE are not aborted, a
client can send them and then quit.

A command whose last reply is being sent is not aborted, there is
nothing left to interrupt. Protocol::end_statement() sets
DISCONNECT_UNINTERRUPTIBLE when it sends the last reply of a command:
an error, or a reply with no more results to follow. A read-only SELECT
sends its result early, from select_send::send_eof(), and the client
can quit while the server still cleans up. COM_STATISTICS writes its
reply itself and sets the flag too. The lost connection is then logged
as usual. The watcher and the connection thread order their accesses
with atomic read-modify-write operations on the flags, no lock is taken
for a command.

A connection that is already being killed is left alone. KILL
CONNECTION shuts the socket of its victim, which the watcher reports,
and a second awake of the dying thread would wait for it in
abort_current_cond_wait() for up to 2 seconds.

The socket is not shut down, the connection ends after the error is
sent, and the reason, "Client disconnected", is logged once. A
binlog dump thread logs "BINLOG DUMP client disconnected", and only
with log_warnings=4: a replica closes the connection on purpose when it
stops. A connection with the disconnect noted, idle or only finishing
its reply, shows that State in the processlist. Galera connections are
not aborted.

The global variable disconnect_watcher, ON by default, can be set OFF,
which also ends the watches that exist. It exists only where there is a
watcher: Linux, Windows, the BSDs and macOS, not the embedded server.

New tests: main.disconnect_*, sys_vars.disconnect_watcher_basic,
rpl.rpl_dump_thread_disconnect and mysql_client_test
test_disconnect_*. Some rpl and innodb tests are adjusted to pass with
the watcher ON and OFF.

Assisted-by: Claude:claude-sonnet-5-5
6236d5052146...
6236d5052146c755f5db624d7f7463c752673653
Hemant Dangi
MDEV-42091: --prepare must not trust plugin_dir from backup-my.cnf

Issue:
backup-my.cnf is part of the backup/transfer directory and can arrive over
an untrusted channel (e.g. a Galera SST stream). --prepare honored its
plugin_dir value as-is, letting whoever supplied the backup redirect
plugin loading to a directory of their choosing.

Solution:
Pass the joiner's own, locally-resolved --plugin-dir on the --prepare
command line: it wins over whatever backup-my.cnf says, regardless of
content, unless the user already chose one explicitly via
inno-apply-opts, which takes priority as a deliberate local choice.
Resolve it by asking the joiner's own mariadbd, which is config- and
compiled-default-aware, falling back to a plain config lookup if that's
unavailable. Add a regression test driving the real SST joiner script
directly: deliver a backup stream whose backup-my.cnf claims an
attacker-chosen plugin_dir, and confirm that directory is never
referenced.
5ab6d66f83fc...
5ab6d66f83fc9d9b2fe102e042cedb009be3a6af
Alexey Yurchenko
MDEV-41851 register cache manager with SAVEPOINT handler list

MDEV-41851 happens because by the time SAVEPOINT is called there
exists no connection cache manager yet and it does not care to
create one. As a result the cache manger is lazily created later
but is missing from the SAVEPOINT's handler list and thus is not
fired on ROLLBACK TO SAVEPOINT keeping the event in the cache
which in turn is replicated on subsequent COMMIT and applied on
the replicas.
Make sure to create cache manager in wsrep_register_binlog_handler()
if it does not yet exist.
5a220fbfb504...
5a220fbfb50407e51d5d7ad1087c12a56c92e721
Vladislav Vaintroub
MDEV-38741 Abort the running statement when its client disconnects

The server notices that a client is gone only when it reads from or
writes to the socket. A statement that waits, for example for an InnoDB
row lock, keeps running, and its transaction keeps its locks.

Watch the client socket from the end of login until the connection ends
or the client sends COM_QUIT, with epoll (Linux), kqueue (BSD, macOS),
or on Windows with WSAEventSelect and an I/O completion port, and the
exit of the client process for a named pipe. When the client has gone,
the watcher sets DISCONNECT_SIGNALED in THD::disconnect_flags. If a
command that expects an answer is running, it is aborted with
KILL_CONNECTION. If the connection is idle, the next such command is
aborted when it starts. COM_QUIT and COM_STMT_CLOSE are not aborted, a
client can send them and then quit.

A command that is only sending its reply is not aborted, there is
nothing left to interrupt: it sets DISCONNECT_UNINTERRUPTIBLE before
the reply is written, and the lost connection is then logged as usual.
The watcher and the connection thread order their accesses with atomic
read-modify-write operations on the flags, no lock is taken for a
command.

The socket is not shut down, the connection ends after the error is
sent, and the reason, "Client disconnected", is logged once. A
connection with the disconnect noted, idle or only finishing its reply,
shows that State in the processlist. Galera connections are not
aborted.

The global variable disconnect_watcher, ON by default, can be set OFF,
which also ends the watches that exist. It exists only where there is a
watcher: Linux, Windows, the BSDs and macOS, not the embedded server.

New tests: main.disconnect_*, sys_vars.disconnect_watcher_basic and
mysql_client_test test_disconnect_*. Some rpl and innodb tests are
adjusted to pass with the watcher ON and OFF.

Assisted-by: Claude:claude-sonnet-5-5
5381e7b3c52f...
5381e7b3c52fb8a374322b90f3af2b9ee7162f75
Sergei Golubchik
MDEV-40530 Bind the client authentication response to the server certificate

MDEV-31855 introduced ssl_info that let the client verify
that it uses the correct certificate. But MitM could hijack the
connection after the authentication - the client will detect it
and disconnect but the server won't and will continue talking
to the attacker. Let's introduce ssl_info_client packet that
enables the server to verify that the client uses the correct
certificate.

When require_secure_transport=ON and the connection isn't local:
* the server sets MARIADB_CLIENT_SSL_INFO_CLIENT in the initial handshake
* the client sets it in the reply, if it understands it
* when both sides have it, the client sends ssl_info_client right
  after the OK packet, acl_authenticate() reads it before the session
  is used (init_connect, audit, etc)
* a connection is rejected unless ssl_info_client is valid, either
  for the server certificate fingerprint or for "CA" (the client
  verified the certificate itself)
* an account without a password hash (plugin has no hash_password()
  or empty password) is rejected unless the account REQUIRE a client
  certificate

Assisted-by: Claude:claude-opus-5-5
50201501835b...
50201501835be2aef2363bf906071346550c225f
Vladislav Vaintroub
MDEV-38741 Abort the running statement when its client disconnects

The server notices that a client is gone only when it reads from or
writes to the socket. A statement that waits, for example for an InnoDB
row lock, keeps running, and its transaction keeps its locks.

Watch the client socket from the end of login until the connection ends
or the client sends COM_QUIT, with epoll (Linux), kqueue (BSD, macOS),
or on Windows with WSAEventSelect and an I/O completion port, and the
exit of the client process for a named pipe. When the client has gone,
the watcher sets DISCONNECT_SIGNALED in THD::disconnect_flags. If a
command that expects an answer is running, it is aborted with
KILL_CONNECTION. If the connection is idle, the next such command is
aborted when it starts. COM_QUIT and COM_STMT_CLOSE are not aborted, a
client can send them and then quit.

A command whose last reply is being sent is not aborted, there is
nothing left to interrupt. Protocol::end_statement() sets
DISCONNECT_UNINTERRUPTIBLE when it sends the last reply of a command:
an error, or a reply with no more results to follow. A read-only SELECT
sends its result early, from select_send::send_eof(), and the client
can quit while the server still cleans up. The lost connection is then
logged as usual. The watcher and the connection thread order their
accesses with atomic read-modify-write operations on the flags, no lock
is taken for a command.

A connection that is already being killed is left alone. KILL
CONNECTION shuts the socket of its victim, which the watcher reports,
and a second awake of the dying thread would wait for it in
abort_current_cond_wait() for up to 2 seconds.

The socket is not shut down, the connection ends after the error is
sent, and the reason, "Client disconnected", is logged once. A
connection with the disconnect noted, idle or only finishing its reply,
shows that State in the processlist. Galera connections are not
aborted.

The global variable disconnect_watcher, ON by default, can be set OFF,
which also ends the watches that exist. It exists only where there is a
watcher: Linux, Windows, the BSDs and macOS, not the embedded server.

The global suppressions of "Aborted connection" in mtr no longer hide
the reason: they hid the aborts of ordinary clients. The tests that
drop a client on purpose accept it, through have_disconnect_watcher.inc,
dirty_close and mysql_client_test.

New tests: main.disconnect_*, sys_vars.disconnect_watcher_basic and
mysql_client_test test_disconnect_*. Some rpl and innodb tests are
adjusted to pass with the watcher ON and OFF.

Assisted-by: Claude:claude-sonnet-5-5
4ccee172abe5...
4ccee172abe5ae923c597e6c7700acd9d61df2f2
Monty
MDEV-41811 Atomic CoR: a failed CREATE OR REPLACE ... SELECT NEXTVAL() loses the old table

The reason for this failure is commit
bb4dd70e7c6509ed9bf8c38d44dffb6d2fcca415 which generated a new error code
in ha_discover_table() that ddl_log_revert() for create and replace table
cannot handle. The code also did not take into account that share and
thd->lex->query_tables are not guaranteed to refer to the same table.

I removed the code for handling sequences from ha_discover_table() and did
instead generate the correct error in open_table_error(). This handles
many cases the old code did not handle.
48deaea65d29...
48deaea65d29514a6eb23b3f2ae67a201f160dd5
Oleksandr Byelkin
Merge branch 'br-11.8-merge' into br-12.3-merge
3e0aa429c5a5...
3e0aa429c5a55861daab12fd221938e264d6682c
Oleksandr Byelkin
fix
31cd3e717a11...
31cd3e717a1108b900f16974f5023f706515a7ec
Monty
Compiling ha_videx statically causes 'drop table xxx' to not generate errors

The issue was that ha_videx::delete_table() returns 0, which means that
the table was successfully deleted, which confused delete_table_force to
belive that the table was succesfully deleted, which in turn casued
the "Unknown table" error to not be given.
2cd4ffc292cc...
2cd4ffc292cccbedc9e726c77ff5ac0d3e228a1c
Monty
MDEV-41811 Atomic CoR: a failed CREATE OR REPLACE ... SELECT NEXTVAL() loses the old table

The reason for this failure is commit
bb4dd70e7c6509ed9bf8c38d44dffb6d2fcca415 (MDEV-13005) which generated
a new error code in ha_discover_table() that ddl_log_revert() for
create and replace table cannot handle. The code also did not take
into account that share and thd->lex->query_tables are not guaranteed
to refer to the same table.

I removed the code for handling sequences from ha_discover_table() and
instead generate the correct error in open_table_error(). This
handles many cases the old code did not handle.

Changes:
- share->table_type == TABLE_TYPE_SEQUENCE is set early in open for
  sequences to be able to generate correct error messages
- share->table_type == TABLE_TYPE_VIEW is now set for views during open
  This is mainly to make the table_type have an usable value for views.
26c094ce1c28...
26c094ce1c286813f89ab6cfad5cddb763b6b591
Sergei Golubchik
MDEV-40530 require_secure_transport=CLIENT: the client says it has verified

With require_secure_transport=CLIENT a client without a password hash
must still set MARIADB_CLIENT_SSL_INFO_CLIENT and say that it has
verified the server certificate: ssl_info_client is COM_QUIT and the
version if it has, and COM_QUIT alone if it has not.

This is not a proof, a man in the middle can say it too, but a client
that does not verify the server certificate is rejected and an old
client that does not know ssl_info_client is rejected too.

A client cannot simply abort after the OK packet if it has not
verified the certificate: an account that requires a client
certificate does not need ssl_info_client.

Assisted-by: Claude:claude-opus-5-5
206632d1bc5b...
206632d1bc5bcdf7469a5e893a0642e8bf5f6973
Hemant Dangi
MDEV-41158: authenticate the mariabackup SST backup-data stream too

Issue:
The joiner's per-transfer secret was only generated for
ssl-mode=VERIFY_*, and checked only on the small GTID stage: the
backup-data stage itself was accepted unauthenticated, letting any
TCP peer reaching the SST port supply a backup --prepare then runs.

Solution:
Generate the secret unconditionally and require it on the backup-data
stream too, built in lockstep with the existing encrypt/decompress/
progress wrapping so the check can't be silently skipped by it. Add
sst-mariabackup-allow-unauthenticated-donor for rolling upgrades,
failing closed on invalid values; VERIFY_*'s own pre-existing secret
check is unaffected by it. Add a regression test covering the
authenticated pipeline with that wrapping active.
170c09b7d61c...
170c09b7d61c9223079ce4ec95af693b3df8e002
Alexey (Holyfoot) Botchkov
MDEV-41821 XMLISVALID returns unexepected result, various corner cases.

Control of empty builtin types added.
complexType with only optional particles allows empty element.
maxOccurs now handled in <choice>.
No type in <element> now is allowed.
0db6d3841fe3...
0db6d3841fe324d62f0b932efbcd0f3623c86b34
Oleksandr Byelkin
MDEV-38274 fix lost in 11.4->11.8 merge

The merge 0e86a0d055f kept the old 11.8 version of
main.analyze_stmt_prefetch_count. Take the 11.4 version and keep
"character set latin1" from MDEV-19123.

Assisted-by: Claude:claude-opus-5-5
0c45e8a5f1b8...
0c45e8a5f1b8492c36d0b9ecc6ff302831a5c3fd
Daniel Black
oqgraph: clang 23 warning unused-but-set-global

Removed the oqgraph_init_done and the fini function
as setting this variable was the only thing it did.

Resolve complile error:

oqgraph/ha_oqgraph.cc:131:13: error: variable 'oqgraph_init_done' set but not used [-Werror,-Wunused-but-set-global]
  131 | static bool oqgraph_init_done= 0;
0a0766bafee4...
0a0766bafee43c407720cf30cd4d69d8c63673e3
Monty
Enabling ha_videx causes 'drop table xxx' to not generate errors

Fixed by having videx_hton->drop_table return -1 (no table data attached
to this object)
BuildBot (0.8.9) working for the MariaDB project.
Page built: Mon 09 Nov 2026 01:04:19 (UTC)