- 10 Mar, 2018 4 commits
-
-
Marko Mäkelä authored
fil_space_t::atomic_write_supported: Always set this flag for TEMPORARY TABLESPACE and during IMPORT TABLESPACE. The page writes during these operations are by definition not crash-safe because they are not written to the redo log. fil_space_t::use_doublewrite(): Determine if doublewrite should be used. buf_dblwr_update(): Add assertions, and let the caller check whether doublewrite buffering is desired. buf_flush_write_block_low(): Disable the doublewrite buffer for the temporary tablespace and for IMPORT TABLESPACE. fil_space_set_imported(), fil_node_open_file(), fil_space_create(): Initialize or revise the space->atomic_write_supported flag. buf_page_io_complete(), buf_flush_write_complete(): Add the parameter dblwr, to indicate whether doublewrite was used for writes. buf_dblwr_sync_datafiles(): Remove an unnecessary flush of persistent tablespaces when flushing temporary tablespaces. (Move the call to buf_dblwr_flush_buffered_writes().)
-
Marko Mäkelä authored
buf_flush_init_for_writing(): Remove the parameter skip_checksum.
-
Marko Mäkelä authored
fsp_init_file_page_low(): Always initialize the page.
-
Marko Mäkelä authored
-
- 08 Mar, 2018 1 commit
-
-
Jan Lindström authored
There were some differences on MariaDB compared with MySQL. Fix these differences.
-
- 07 Mar, 2018 1 commit
-
-
Ian Gilfillan authored
-
- 02 Mar, 2018 1 commit
-
-
Elena Stepanova authored
-
- 25 Feb, 2018 1 commit
-
-
Alexander Barkov authored
MDEV-15420 Wrong result for CAST from TIME or DATETIME with zero integer part and non-zero microseconds to DECIMAL(X,Y) The loop in ull2dec() does not iterate if "from" is zero, so to->intg got erroneously set to 0 instead of 1. Because if this, my_decimal2seconds() wrote the fractional part into a wrong buf[x]. Catching the special case with zero "from" and properly initialize "to" using decimal_make_zero().
-
- 24 Feb, 2018 1 commit
-
-
Jan Lindström authored
MDEV-14814: encryption.innodb_encryption-page-compression failed in buildbot with timeout on wait condition Fix wrong wait condition.
-
- 22 Feb, 2018 7 commits
-
-
Sergei Golubchik authored
fix the compilation error. no support for plugins yet.
-
Sergei Golubchik authored
-
Sergei Golubchik authored
-
Jan Lindström authored
-
Sergei Golubchik authored
-
Sergei Golubchik authored
rocksdb and spider
-
Sergei Golubchik authored
-
- 21 Feb, 2018 8 commits
-
-
Sergei Golubchik authored
despite the name, it logs both stdout and stderr
-
Sergei Golubchik authored
tokudb apparently requires perfschema now
-
Sergei Golubchik authored
-
Sergei Petrunia authored
-
Daniel Black authored
Analysis:- The problem is the change in the implementation of wait_for_listen in wsrep_sst_xtrabackup-v2.sh. The new script uses lsof which will always exit with an error code if it can't find all the items, and because the script has the -e option set in the hashbang line (#!/bin/bash -ue), the script will abort right after running lsof if lsof can't find even a single item among all the items listed in its arguments. This will happen even if socat is running and listening, because it can't find nc. The loop in wait_for_listen will therefore always quit after one iteration without writing the "ready" line to signal the parent. Solution:- We will or the lsof with true. Patch Credit :Daniel Black and David Wang
-
Sergei Petrunia authored
Part #2: some transactions have m_rocksdb_tx==NULL (and most functions of Rdb_transction_impl handle this case. Do like they do)
-
Sergei Golubchik authored
-
Sergei Petrunia authored
Make MyRocks' non-XA commit path to first do the commit without syncing and then sync.
-
- 20 Feb, 2018 3 commits
-
-
Vladislav Vaintroub authored
a) We do not need this on Windows, and it is not clear what it does,inside service. b) It hinders debugging. mysql-test-run.pl --debugger=vsjitdebugger more often than , would stop here throwing "invalid handle" exception.
-
Jan Lindström authored
* galera_gtid_slave * galera_unicode_identifiers
-
Jan Lindström authored
-
- 19 Feb, 2018 9 commits
-
-
Galina Shalygina authored
the non-recursive CTE via prepared statement The problem appears as the column names of the CTE were allocated on the wrong MEMROOT and after the preparation of the statement they disappear. To fix it in the procedure With_element::rename_columns_of_derived_unit the CTE column names are now allocated in the permanent MEMROOT for the prepared statements and stored procedures.
-
Vladislav Vaintroub authored
This reverts commit 0ea45725, since it breaks clang
-
Vladislav Vaintroub authored
-
Marko Mäkelä authored
fsp_fill_free_list(): Correctly determine whether the temporary tablespace file should be extended in order to respond to a page allocation request. The inverted condition was noticed by Thiru when he analyzed MDEV-13013.
-
Sergey Vojtovich authored
Added more files affected by GCC ICE: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67143
-
Sergey Vojtovich authored
Added missing include.
-
Marko Mäkelä authored
-
Marko Mäkelä authored
srv_conc_t::n_active: Correct the comment, and remove an assertion that trivially holds now that the type is unsigned.
-
Jan Lindström authored
MDEV-14814: encryption.innodb_encryption-page-compression failed in buildbot with timeout on wait condition Test changes only.
-
- 18 Feb, 2018 3 commits
-
-
Vladislav Vaintroub authored
-
Vladislav Vaintroub authored
-
Jan Lindström authored
-
- 17 Feb, 2018 1 commit
-
-
Jan Lindström authored
pages.
-