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
Mohamed
MDEV-40423 use-of-uninitialized-value in Gis_point::get_data_as_wkt(String*, char const**) const

read_point_from_json() returned success for a GeoJSON position with
fewer than two numbers, leaving x and/or y uninitialized. They were
then stored in the WKB and read by ST_AsText().

Reject such positions as incorrect GeoJSON, since RFC 7946 requires
at least two numbers.
Mohamed
MDEV-40520 SHOW STATUS triggers MSAN use-of-uninitialized-value in get_one_variable()

init_simple_key_cache() resets the key cache statistics counters before
building the cache, except warm_blocks, which was set only when the
cache got at least 8 blocks. With a key_buffer_size too small for that,
such as 0, warm_blocks kept the uninitialized value from my_malloc() and
SHOW STATUS printed it as Key_blocks_warm.

Reset warm_blocks together with the other counters.
Oleksandr Byelkin
Merge branch '13.1' into 13.2
Oleksandr Byelkin
Merge branch 'bb-13.1-release' into bb-13.2-release
Monty
Making LOCK TABLES work with GTT tables as it works with TEMPORARY tables.

Changes:
- A GTT instance doesn't keep a metadata lock on the definition. The lock
  is released as soon as the instance is created, and a second use in the
  same statement reuses the instance without taking a lock. Because of
  this, ALTER or DROP of the definition from another session no longer
  waits.

- LOCK TABLES for GTT always uses the GTT instance, not the table
  definition. One side effect of this is that one cannot do a DDL on
  the GTT definition, like ALTER, RENAME or DROP while one has a LOCK
  on the table.
- One can use GTT instances without having them in LOCK TABLE, like one
  can do with temporary tables.
- CREATE OR REPLACE GTT over a table locked with LOCK TABLES (plain, LIKE
  and SELECT) releases the replaced table's locks. If it was the last
  locked table, LOCK TABLES ends.
- FLUSH TABLES now ignores temporary tables and GTT instances under LOCK
  TABLES. Without LOCK TABLES, FLUSH TABLE of a GTT removes the
  definition from the table cache, as before.
- After FLUSH TABLES gtt WITH READ LOCK, statements skip the locked
  definition and use the instance. This fixes an assertion.
- DDL on a GTT definition under LOCK TABLES fails:
  CREATE OR REPLACE (all kinds) gives ER_TABLE_NOT_LOCKED.
  ALTER, RENAME and DROP give ER_TABLE_IN_USE while the session has a
  GTT instance, and ER_TABLE_NOT_LOCKED when it has none (for example
  after TRUNCATE).
- SHOW CREATE TABLE shows the session's instance if there is one. Under
  LOCK TABLES it can also show a definition that has no instance.
- "LOCK TABLE gtt_table write" now succeeds inside a read-only transaction
- mariadb-dump will not try to dump content of GTT tables.
- Moved MDL_context::release_lock(enum_mdl_duration duration,
  MDL_ticket *ticket) from private to public and added a DBUG_ASSERT if
  the duration is incorrect.
- The pseudo_slave_mode test cases now fail at LOCK TABLES.
- Removed defines/functions:  OPEN_FOR_LOCKED_TABLES_LIST,
  select_create::parent_global_tmp_table
- CREATE OR REPLACE under LOCK TABLES now releases the locks of a dropped
  table after binary logging, also for normal tables."

Bugs fixed:
- A failed CREATE GLOBAL TEMPORARY TABLE ... SELECT left the GTT
  definition behind (on disk, or in the table definition cache if the
  GTT instance could not be opened).
Sergei Golubchik
13.3 branch
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 with mtr.add_suppression(),
details in PR#6035.

Assisted-by: Claude:claude-sonnet-5-5
Vladislav Vaintroub
MDEV-38741 Abort the running statement when its client disconnects

A watcher thread subscribes to the OS notification that a client has
gone (e.g. TCP FIN or RST for a TCP socket, a gone client process for a
named pipe) and aborts the running statement like KILL CONNECTION does.
Controlled by the global variable disconnect_watcher. Details in
PR#6035.

Assisted-by: Claude:claude-sonnet-5-5
Vladislav Vaintroub
MDEV-42258: Windows ignores --back-log, simultaneous connects are refused

On Windows, listen() with a positive backlog is limited to a small value,
so a burst of simultaneous connections (e.g. 256 clients) overflows the
accept queue and some are refused with WSAECONNREFUSED (10061).

For TCP listening sockets, including the extra port, pass
SOMAXCONN_HINT(back_log) when back_log >= 200, the range it supports
(200-65535). Smaller values are passed as before.

Assisted-by: Claude:claude-sonnet-5-5
Oleksandr Byelkin
do not leave sleep behind.
Oleksandr Byelkin
Merge branch 'main' into bb-13.2-release
Vladislav Vaintroub
MDEV-38741 Abort the running statement when its client disconnects

A watcher thread subscribes to the OS notification that a client has
gone (e.g. TCP FIN or RST for a TCP socket, a gone client process for a
named pipe) and aborts the running statement like KILL CONNECTION does.
Controlled by the global variable disconnect_watcher. Details in
PR#6035.

Assisted-by: Claude:claude-sonnet-5-5
Vladislav Vaintroub
MDEV-42049 MTR server-cert.pem has no DNS SAN, fails on OpenSSL 4.1

OpenSSL 4.1 no longer matches a host name against the subject CN when
the certificate has no DNS subjectAltName. Re-sign server-cert.pem with
subjectAltName=DNS:localhost, and add it to generate-ssl-certs.sh.

main.ssl_7937, main.ssl_system_ca and rpl.rpl_ssl1 fail with
OpenSSL 4.1 without this change.

The SHA1 fingerprint changes. main.ssl_fp (11.4+) must be updated when
merging up: F1:D0:08:AF:A1:D2:F4:15:79:B4:39:06:41:F4:20:96:F1:90:A9:65
-> AE:F0:29:15:F1:9E:A0:A0:69:FB:83:20:B9:30:E3:DE:2F:25:2C:37

Assisted-by: Claude:claude-sonnet-5-5
Tarun Wadhwa
MDEV-37605: Extend mariadb-binlog to convert InnoDB-based binlogs to legacy format

MariaDB 12.3 added support for storing binary logs inside InnoDB
tablespace files (.ibb) instead of traditional flat sequential files.
There was previously no way to convert these back to the legacy
format, which is needed for downgrades, migrations, and third-party
tooling that only understands classic binlogs.

This adds a --convert-engine-binlog option to mariadb-binlog that
reads one or more InnoDB-based binlogs and writes out equivalent
legacy-format binlog files.

Usage:
  mariadb-binlog --convert-engine-binlog -r output_legacy_binlog \
      input1.ibb input2.ibb

This produces output_legacy_binlog.0000001, .0000002, etc. Note
there isn't a direct correlation between the number of input files
and output files. The output is split whenever the current file
hits --max-binlog-size (default 1GB, since legacy binlogs cannot
exceed 4GB), and also on every server restart recorded in the
InnoDB binlog, so a single input file can produce several output
files and a restart boundary can force a split even before the
size limit is reached.

Additionally, --max-binlog-size is also supported as a standalone
option to control this rotation threshold directly.

Only --convert-engine-binlog, -r, and --max-binlog-size are
supported alongside this mode; any other option will cause
mariadb-binlog to throw an error

Caveats:
- Encryption: InnoDB-based binlogs don't currently support
  encryption, so the converter doesn't handle it either. Output
  legacy binlogs are never encrypted by this mode.
- Checksums: InnoDB-based binlogs rely on page-level checksums
  instead of the legacy per-event checksums, so the initial
  implementation generates legacy output with no event-level
  checksum.

Reviewed-by: Kristian Nielsen <[email protected]>
Reviewed-by: Brandon Nesterenko <[email protected]>
ParadoxV5
forgot to update this test
ParadoxV5
missed this spot
Monty
squash! 02490ed2c6c26194f84889e96e17a0536bb0275e

- Moved GTT variables HA_OPTION_.. variables to an enum in handler.cc:
  GTT_NONE=0, GTT_TABLE=1, GTT_ON_COMMIT_DELETE_ROWS=2. The gtt options
  are now stored in TABLE_SHARE->gtt_options and TABLE_LIST->gtt_options.
- Changed class Tmp_table_kind to an enum
  {TMP_TABLE_LOCAL, TMP_TABLE_GLOBAL, TMP_TABLE_LOCAL_OR_GLOBAL }
- Moved find_tmp_table_share_w_base_key() to use
  find_tmp_table_share(..., 1). This was because functions where
  practically identical, except for a length subtraction in one place.
- Fixed wrong test cases
- Fixed compiler errors by moving code from 'main'
Alexey Yurchenko
MGL-301 Add MTR tests and fix error reporting
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 with mtr.add_suppression(),
details in PR#6035.

Assisted-by: Claude:claude-sonnet-5-5
forkfun
MDEV-42222 String copy must not share a heap buffer

A copy of a String pointed to the buffer of the original and kept its
capacity without owning the buffer. Condition pushdown copies items
with shallow_copy(): all String members of the copy, and
Arg_comparator::value1/value2 of a copied comparison, shared the
buffers of the original item. When the original item reallocated
a buffer for a longer value, the copy kept using the freed one,
e.g. UPPER, CAST(.. AS CHAR), STR_TO_DATE and JSON_EXTRACT
comparison on 2nd execution of a prepared statement.

The copy constructor of a String that owns its buffer copies the data
now. A copy of a String that doesn't own its buffer points to the same
data and has no capacity.

Assignment still makes the String point to the data of the other one,
but without capacity. No code that writes through an assigned String
was found.

SFORMAT copied every String argument twice per row. It passes them by
reference now, to avoid copying the data.

Tests: main.in_subq_cond_pushdown and main.ctype_ucs.

Assisted-By: Claude:claude-sonnet-5-5
Vladislav Vaintroub
MDEV-42258: Windows ignores --back-log, simultaneous connects are refused

On Windows, listen() with a positive backlog is limited to a small value,
so a burst of simultaneous connections (e.g. 256 clients) overflows the
accept queue and some are refused with WSAECONNREFUSED (10061).

For TCP listening sockets, including the extra port, pass
SOMAXCONN_HINT(back_log) when back_log >= 200, the range it supports
(200-65535). Smaller values are passed as before.

Assisted-by: Claude:claude-sonnet-5-5
Vladislav Vaintroub
MDEV-40382 Support --ssl-crl on server builds with WolfSSL

--ssl-crl was silently discarded on WolfSSL: sslopt-case.h nulled
opt_ssl_crl after option parsing, and new_VioSSLFd() skipped CRL
loading for HAVE_WOLFSSL. A revoked certificate could still
authenticate a REQUIRE X509 / REQUIRE SUBJECT account.

WolfSSL's OpenSSL-compat layer already implements
X509_STORE_load_locations/X509_STORE_set_flags with real CRL
enforcement, so remove both restrictions.

Also fix X509_STORE_set_flags()'s success check: unlike OpenSSL,
WolfSSL can return negative error codes on failure, which == 0
missed.

Existing CRL tests were fixed to run with WolfSSL as well.

Note: on Windows, a revoked cert sometimes surfaces as ECONNRESET
instead of a TLS alert; the test's regex handles this too.
Monty
Rename gtt functions.

Renames:
- global_tmp_table() -> is_gtt()
- local_tmp_table()  -> is_local_tmp_table()

New functions:
- is_gtt_instance()
- is_gtt_definition()
Vladislav Vaintroub
MDEV-38741 mariabackup: run the debug DDL as one statement

The debug thread of mariabackup that emulates a DDL sends several
statements in one query, and closes its connection as soon as the first
of them returned. The server used to run the rest. A server that aborts
the statement of a client that is gone does not, and
mariabackup.innodb_ddl_on_intermediate_table hangs, as the debug sync
signals the main thread waits for do not come. Run the statements in
one BEGIN NOT ATOMIC block, so that the client waits for all of them.
The ALTER of that test fails with a lock wait timeout on purpose, which
the handler of the block ignores, as the thread did before.

Assisted-by: Claude:claude-sonnet-5-5
Sergei Golubchik
remove two deprecated hashicorp_key_management_* variables

hashicorp_key_management_use_cache_on_timeout
- was default to enabled, now enabled always

hashicorp_key_management_cache_timeout
- was default to INT64_MAX, now no timeout
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 with mtr.add_suppression(),
details in PR#6035.

Assisted-by: Claude:claude-sonnet-5-5
Monty
Removed mdl_ticket from TMP_TABLES_SHARE

This is preparation for versioning of GTT tables and also remove
LOCK TABLE for GTT instances and make GTT tables work as normal temporary
Daniel Black
MDEV-41860 return vector extended metadata

This adds "vector" as an extended metadata type
for when is returning a collection of 4 byte little
endian floats from the server. This applies to the
data type and also the vec_fromtext function.

Guidance thanks to Vivaan Sanghavi.

Test case/review: Sergei Golubchik
Monty
Changed ha_table_exists() to return Table_type to provide more information

- Added TABLE_TYPE_GLOBAL_TEMPORARY to Table_type
- Added TABLE_TYPE_FRM_ONLY to allow ha_table_exists() to inform caller
  that frm file exists but is not readable or has bad information.
- Removed parameter is_sequence from ha_table_exists()
ParadoxV5
MDEV-21322 Report Slave Progress to the Master

(Draft: Commit message is staged on the PR.)
Vladislav Vaintroub
MDEV-42049 MTR server-cert.pem has no DNS SAN, fails on OpenSSL 4.1

OpenSSL 4.1 no longer matches a host name against the subject CN when
the certificate has no DNS subjectAltName. Re-sign server-cert.pem with
subjectAltName=DNS:localhost, and add it to generate-ssl-certs.sh.

main.ssl_7937, main.ssl_system_ca and rpl.rpl_ssl1 fail with
OpenSSL 4.1 without this change.

The SHA1 fingerprint changes. main.ssl_fp (11.4+) must be updated when
merging up: F1:D0:08:AF:A1:D2:F4:15:79:B4:39:06:41:F4:20:96:F1:90:A9:65
-> AE:F0:29:15:F1:9E:A0:A0:69:FB:83:20:B9:30:E3:DE:2F:25:2C:37

Assisted-by: Claude:claude-sonnet-5-5
Monty
Release the GTT metadata locks at the end of open_tables()
Sergei Golubchik
remove two hashicorp_key_management_* variables

hashicorp_key_management_use_cache_on_timeout
- was default to enabled, now enabled always

hashicorp_key_management_cache_timeout
- was default to INT64_MAX, now no timeout
ParadoxV5
MDEV-21322 Report Slave Progress to the Master

(Draft: Commit message is staged on the PR.)
Oleksandr Byelkin
Merge branch '12.3' into 13.1
Tarun Wadhwa
MDEV-37605: Extend mariadb-binlog to convert InnoDB-based binlogs to legacy format

MariaDB 12.3 added support for storing binary logs inside InnoDB
tablespace files (.ibb) instead of traditional flat sequential files.
There was previously no way to convert these back to the legacy
format, which is needed for downgrades, migrations, and third-party
tooling that only understands classic binlogs.

This adds a --convert-engine-binlog option to mariadb-binlog that
reads one or more InnoDB-based binlogs and writes out equivalent
legacy-format binlog files.

Usage:
  mariadb-binlog --convert-engine-binlog -r output_legacy_binlog \
      input1.ibb input2.ibb

This produces output_legacy_binlog.0000001, .0000002, etc. Note
there isn't a direct correlation between the number of input files
and output files. The output is split whenever the current file
hits --max-binlog-size (default 1GB, since legacy binlogs cannot
exceed 4GB), and also on every server restart recorded in the
InnoDB binlog, so a single input file can produce several output
files and a restart boundary can force a split even before the
size limit is reached.

Additionally, --max-binlog-size is also supported as a standalone
option to control this rotation threshold directly.

Only --convert-engine-binlog, -r, and --max-binlog-size are
supported alongside this mode; any other option will cause
mariadb-binlog to throw an error.

Caveats:
- Encryption: InnoDB-based binlogs don't currently support
  encryption, so the converter doesn't handle it either. Output
  legacy binlogs are never encrypted by this mode.
- Checksums: InnoDB-based binlogs rely on page-level checksums
  instead of the legacy per-event checksums, so the initial
  implementation generates legacy output with no event-level
  checksum.

Reviewed-by: Kristian Nielsen <[email protected]>
Reviewed-by: Brandon Nesterenko <[email protected]>
Vladislav Vaintroub
MDEV-38741 Abort the running statement when its client disconnects

A watcher thread subscribes to the OS notification that a client has
gone (e.g. TCP FIN or RST for a TCP socket, a gone client process for a
named pipe) and aborts the running statement like KILL CONNECTION does.
Controlled by the global variable disconnect_watcher. Details in
PR#6035.

Assisted-by: Claude:claude-sonnet-5-5
ParadoxV5
MDEV-21322: Delete extra `remove_slave()` call

The `Ack_receiver::remove_slave()` call in
`kill_zombie_dump_threads()`’s loop callback is not necessary,
because binlog dump threads that have called `Ack_receiver::add_slave()`
will call `remove_slave()` during cleanup anyway.
This commit erases this extraneous call, for it wastes the dump thread’s
normal `remove_slave()` step, which locks `Ack_receiver`’s global mutex.
MDEV-21322 (report ACKs on SHOW SLAVE HOSTS) also has trouble
matching its synchronization needs with this global lock acquisition.