1
0
mirror of https://github.com/mariadb-corporation/mariadb-columnstore-engine.git synced 2025-11-22 20:42:36 +03:00
Commit Graph

109 Commits

Author SHA1 Message Date
Roman Nozdrin
cfce821513 MCOL-4337 Controllernode now tries to establish connections to
all workernodes when initialising
There is a new corresponding XML config options
DBRM_Controller.WorkerConnectionTimeout that controls how long
controllernode waits for wns before it proceeds
2020-10-09 09:12:59 +00:00
David Hall
951a6abe85 MCOL-4330 dev Refresh shared memory
If the plugin is restarted after the system is running, it's possible to have stale references to shared memory. This patch deletes those references so the next time they're used, just-in-time initialization will get the latest.
2020-10-01 18:19:03 -05:00
Roman Nozdrin
7e868bc588 add dependency for generated header files errorids.h messageids.h patch made by Ben. 2020-07-10 18:38:49 +00:00
benthompson15
eac7dab096 MCOL-4030: first commit of warning removals unneed const and missing virtual dtors. 2020-06-23 13:51:36 -05:00
David Hall
1d5e820e97 Merge branch 'develop' into MCOL-3536
Conflicts:
	CMakeLists.txt
2020-06-04 16:22:59 -05:00
Patrick LeBlanc
5440977f45 Merge pull request #1197 from tntnatbry/develop-batchinserts-improve
Improve batch inserts.
2020-06-02 14:05:47 -05:00
David Hall
06e50e0926 MCOL-3536 collation 2020-05-26 12:42:11 -05:00
David Hall
8479a87e46 Merge branch 'develop' into MCOL-3536 2020-05-18 16:22:01 -05:00
David Hall
1f3d1e6fd6 MCOL-3536 collation 2020-05-14 16:02:49 -05:00
Jose
98abf95eae MCOL-3991 MCS is now single package and properly uninstalls 2020-05-12 13:36:24 +00:00
Gagan Goel
d4d0ebdf5d Improve batch inserts.
1) Instead of making dbrm calls to writeVBEntry() per block,
     we make these calls per batch. This can have non-trivial
     reductions in the overhead of these calls if the batch size
     is large.

  2) In dmlproc, do not deserialize the whole insertpackage, which
     consists of the complete record set per column, which would be
     wasteful as we only need some metadata fields from insertpackage
     here. This is only done for batch inserts at the moment, this
     should also be applied to single inserts.
2020-05-10 19:38:06 -04:00
Patrick LeBlanc
cb4228d982 Merge branch 'columnstore-1.4.3' into merge-hotfix-patches-1.5
Conflicts:
	VERSION
	dbcon/mysql/ha_mcs.cpp
	dbcon/mysql/ha_mcs_execplan.cpp
	dbcon/mysql/ha_mcs_impl.cpp
	dbcon/mysql/ha_mcs_pushdown.cpp
	oam/install_scripts/columnstore-post-install.in
	oam/install_scripts/columnstore-pre-uninstall.in
	oam/install_scripts/columnstore.in
	oam/install_scripts/post-mysql-install
2020-04-27 17:17:55 -04:00
Roman Nozdrin
6f3ae1b429 MCOL-3941 Ninja now builds MCS.
Generated files in utils/loggingcpp/ are now moved into bin dir.
2020-04-20 13:35:30 +00:00
benthompson15
5c977f2c85 MCOL-3905: slavecomm should use the path provided to the function for openning journal file with the change that added UUID to path for downloaded files during getDBRMData. 2020-04-03 12:41:06 -05:00
benthompson15
accfe3b070 MCOL-3842: we need to delete instances not just release them when a failover happens because the worknodes have changed in configuration files. 2020-04-03 12:17:18 -05:00
Patrick LeBlanc
180140303e MCOL-3829: CS not starting correctly if rebooted at specific point
Found that the IDBDataFile path in BRM journal writing code needs to seek
to the end of the file before writing.

If save_brm is run (as it is during init), it truncates the journal file,
but the workernode retained its original offset, resulting in the front
of the file being 0-filled on the next journal write.  Load_brm can't
load that.
2020-02-24 10:35:34 -05:00
Patrick LeBlanc
295ba65724 MCOL-3829: CS not starting correctly if rebooted at specific point
Found that the IDBDataFile path in BRM journal writing code needs to seek
to the end of the file before writing.

If save_brm is run (as it is during init), it truncates the journal file,
but the workernode retained its original offset, resulting in the front
of the file being 0-filled on the next journal write.  Load_brm can't
load that.
2020-02-24 10:11:55 -05:00
Sergei Golubchik
586391e1ca compilation failure
error: reference to 'mutex' is ambiguous
note: candidates are: 'class boost::mutex'
note:                 'class std::mutex'
2019-12-19 18:13:39 +01:00
Andrew Hutchings
49994f7bc3 Fix warnings found in DEBUG combined build
Fixes:
* Irrelevant where conditions
* Irrelevant const
* A potential infinite loop in treenode
* Bad implicit case fallthroughs
* Explicit markings for required case fallthroughs
* Unused variables
* Unused function

Also disabled some warnings for now which we should fix later.
2019-12-10 16:33:08 +00:00
Andrew Hutchings
7489d0bfd0 MCOL-3625 Rename packages
Rename packages to MariaDB-columnstore-engine, MariaDB-columnstore-libs
and MariaDB-columnstore-platform.

Also add the "columnstore-" prefix the the components so that MariaDB's
packaging system understands then and add a line to include them in
MariaDB's packaging.

In addition
* Fix S3 building for dist source build
* Fix Debian 10 dependency issue
* Fix git handling for dist builds
* Add support for MariaDB's RPM building
* Use MariaDB's PCRE and readline
* Removes a few dead files
* Fix Boost noncopyable includes
2019-12-04 11:04:39 +00:00
Andrew Hutchings
a8cd34f86d Add support for building from server 2019-12-03 21:21:28 +00:00
Andrew Hutchings
8ab9ebb0f4 MCOL-3606 Make ColumnStore use generic paths
ColumnStore now uses standard bin/lib paths for pretty much everything.
Data path is now hard-coded to /var/lib/columnstore.

This patch also:

* Removes v1 decompression
* Removes a bunch of unneeded files
* Removes COLUMNSTORE_INSTALL_DIR / $INSTALLDIR
* Makes my.cnf.d work for all platforms (MCOL-3558)
* Changes configcpp to use recursive mutex (fixes possible config write deadlock)
* Fixes MCOL-3599 Fix regr functions, The library was installed in the wrong location
* Fixes a bunch of Ubuntu packaging issues
* Changes the binary names of several of the executables so as not to
clash with potential executables from other packages
2019-11-09 16:53:05 +00:00
Andrew Hutchings
3fef0f21d3 Remove vpj files
They shouldn't be here
2019-09-05 17:38:03 +01:00
Patrick LeBlanc
fbdf1a5308 Merge branch 'develop' into S3-project 2019-08-14 15:37:14 -05:00
Roman Nozdrin
b1bc995420 Merge branch 'develop' into remove-infinidb 2019-08-13 12:32:01 +03:00
Andrew Hutchings
9d83b49fca MCOL-104 First pass of InfiniDB rename in code 2019-08-12 09:41:28 +01:00
Patrick LeBlanc
a09a9d5d0f Mass substitution 'Corporaton' -> 'Corporation' 2019-08-07 14:43:25 -05:00
Patrick LeBlanc
62c7ab6480 Fixed a merge error. 2019-08-05 16:35:59 -05:00
Patrick LeBlanc
295c638c18 Merge branch 'develop' into S3-project 2019-08-05 16:34:45 -05:00
Patrick LeBlanc
8190e44d1f Fixed an old bug in the HDFS code in the BRM. Journal file supposed
to be opened for appending.
2019-07-29 14:26:13 -05:00
Patrick LeBlanc
a84e3a408f Last commit of the initial set of changes. 2019-07-12 10:52:09 -05:00
Patrick LeBlanc
9ced374c81 Another incremental commit. 2019-07-12 10:33:51 -05:00
Patrick LeBlanc
16dc887f20 Incremental commit. Stripping out the OBE fstream IO code. 2019-07-12 10:06:56 -05:00
Patrick LeBlanc
7dca5365b6 Found other places in the BRM code that were using fstreams vs IDB*. 2019-06-24 16:42:28 -05:00
Patrick LeBlanc
b30ee07318 Merge remote-tracking branch 'upstream/develop' into S3-project 2019-06-24 15:41:10 -05:00
Patrick LeBlanc
157d1ef667 Made the BRM not save a snapshot after every transaction. 2019-06-14 12:33:12 -05:00
Patrick LeBlanc
bcdd76adc0 Fixed a bug in the new extentmap saving code. 2019-06-14 11:50:47 -05:00
Patrick LeBlanc
029b4a61a8 Made oidserver use the IDB* path, made it write the initial
bitmap in one swipe vs small bits at a time.
2019-06-14 10:09:17 -05:00
Patrick LeBlanc
603cbc9314 First cut of improving IO behavior for BRM load/save 2019-06-13 16:14:00 -05:00
Patrick LeBlanc
3d67df18bf Merge branch 'develop' of github.com:mariadb-corporation/mariadb-columnstore-engine into S3-project 2019-05-17 13:24:09 -05:00
Andrew Hutchings
020b211bb7 Merge branch 'develop-1.2' into develop-merge-up-20190514 2019-05-14 13:58:33 +01:00
Roman Nozdrin
7e2cb05624 MCOL-537 There are no CS-specific warnings building with gcc 8.2. 2019-05-07 16:00:05 +03:00
Roman Nozdrin
9dc33c4e82 Another try to cope with warnings under gcc 8.2. 2019-04-29 11:05:03 +03:00
David Mott
515b93cc3d remove faulty test code 2019-04-29 01:28:54 -05:00
David Mott
4b9d046c6e Fully resolve potentially ambiguous symbols by removing using namespace statements from headers which have a cascading effect. This causes potential behavior changes when switching to c++11 since symbols can be exported from std and boost while both have been imported into the global namespace. 2019-04-29 01:21:15 -05:00
David Mott
e65f80f493 delete visual c++ project files. cmake can generate these if needed 2019-04-25 23:35:03 -05:00
Patrick LeBlanc
46163c129a Neutered all of the excessive snapshotting that was going on. 2019-04-10 17:53:05 -05:00
Patrick LeBlanc
dfc29fa833 Eventually found the problem with rename(). Was calling
::unlink() instead of our unlink/remove.
2019-04-10 15:01:29 -05:00
Patrick LeBlanc
c0f93d904a Fixed a bug where the BRM would open the journal for reading when
it should be have open for writing.  Not sure how that ever worked.
2019-04-09 17:34:30 -05:00
Patrick LeBlanc
6e6cfdb59b The BRM structs aren't using IDBDataFile consistently.
Testing whether forcing them to use IDBDataFiles breaks anything.
2019-04-09 15:07:00 -05:00