- developers - lists.mariadb.org

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2702)
by knielsen＠knielsen-hq.org 05 May '09

05 May '09

#At lp:maria 2702 knielsen(a)knielsen-hq.org 2009-05-05 [merge] test merge 2

1 0

Re: [Maria-developers] main.pool_of_threads fail - Solaris 10
by Adam M. Dutko 27 Apr '09

27 Apr '09

Toby, You might also want to check this page out: http://askmonty.org/wiki/index.php/MariaDB_versus_MySQL -Adam

1 0

[Maria-developers] main.pool_of_threads fail - Solaris 10
by Toby Thain 27 Apr '09

27 Apr '09

Hi list, I am setting up a buildbot for Solaris 10/amd64 (later SPARC) and I see this test failure on Sun cc (only). gcc builds pass all tests. (Although a patch is required.) Any clues? --Toby bzr branch lp:maria ... BUILD/compile-solaris-amd64-forte && cd mysql-test/ && ./mysql-test-run ... main.connect [ pass ] 16399 worker[1] mysql-test-run: WARNING: Process [mysqld.1 - pid: 495, winpid: 495, exit: 1792] died main.pool_of_threads [ fail ] Test ended at 2009-04-26 15:18:41 090426 22:17:50 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:17:50 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:17:53 InnoDB: Starting shutdown... 090426 22:17:53 InnoDB: Shutdown completed; log sequence number 0 1943684 090426 22:17:53 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete 090426 22:17:54 InnoDB: Started; log sequence number 0 1943684 090426 22:17:54 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:17:54 [Note] Event Scheduler: Loaded 0 events 090426 22:17:54 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution CURRENT_TEST: rpl.rpl_truncate_3innodb 090426 22:18:01 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:01 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:03 InnoDB: Starting shutdown... 090426 22:18:05 InnoDB: Shutdown completed; log sequence number 0 1968339 090426 22:18:05 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: federated.federated 090426 22:18:05 [Note] Plugin 'InnoDB' disabled by command line option 090426 22:18:05 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:18:05 [Note] Event Scheduler: Loaded 0 events 090426 22:18:05 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution CURRENT_TEST: federated.federated_archive CURRENT_TEST: federated.federated_bug_13118 CURRENT_TEST: federated.federated_bug_25714 CURRENT_TEST: federated.federated_server 090426 22:18:17 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:17 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:17 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: federated.federated_innodb InnoDB: The first specified data file ./ibdata1 did not exist: InnoDB: a new database to be created! 090426 22:18:18 InnoDB: Setting file ./ibdata1 size to 10 MB InnoDB: Database physically writes the file full: wait... 090426 22:18:18 InnoDB: Log file ./ib_logfile0 did not exist: new to be created InnoDB: Setting log file ./ib_logfile0 size to 5 MB InnoDB: Database physically writes the file full: wait... 090426 22:18:18 InnoDB: Log file ./ib_logfile1 did not exist: new to be created InnoDB: Setting log file ./ib_logfile1 size to 5 MB InnoDB: Database physically writes the file full: wait... InnoDB: Doublewrite buffer not found: creating new InnoDB: Doublewrite buffer created InnoDB: Creating foreign key constraint system tables InnoDB: Foreign key constraint system tables created 090426 22:18:20 InnoDB: Started; log sequence number 0 0 090426 22:18:20 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:18:20 [Note] Event Scheduler: Loaded 0 events 090426 22:18:20 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution 090426 22:18:20 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:20 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:21 InnoDB: Starting shutdown... 090426 22:18:23 InnoDB: Shutdown completed; log sequence number 0 46409 090426 22:18:23 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: main.connect 090426 22:18:23 [Note] Plugin 'InnoDB' disabled by command line option 090426 22:18:23 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:18:23 [Note] Event Scheduler: Loaded 0 events 090426 22:18:23 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution 090426 22:18:29 [Note] Event Scheduler: scheduler thread started with id 21 090426 22:18:40 [Note] Event Scheduler: Killing the scheduler thread, thread id 21 090426 22:18:40 [Note] Event Scheduler: Waiting for the scheduler thread to reply 090426 22:18:40 [Note] Event Scheduler: Stopped 090426 22:18:40 [Note] Event Scheduler: scheduler thread started with id 77 090426 22:18:40 [Note] Event Scheduler: Killing the scheduler thread, thread id 77 090426 22:18:40 [Note] Event Scheduler: Waiting for the scheduler thread to reply 090426 22:18:40 [Note] Event Scheduler: Stopped 090426 22:18:40 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:40 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:40 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: main.pool_of_threads Unknown option to thread_handling: pool-of-threads Alternatives are: 'one-thread-per-connection','no-threads' 090426 22:18:40 [Note] Plugin 'InnoDB' disabled by command line option 090426 22:18:40 [ERROR] /home/buildbot/maria/sql/mysqld: unknown variable 'thread_pool_size=2' 090426 22:18:40 [ERROR] Aborting 090426 22:18:40 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete Failed to start mysqld.1 - saving '/home/buildbot/maria/mysql-test/var/log/ main.pool_of_threads/' to '/home/buildbot/maria/mysql-test/var/log/ main.pool_of_threads/' Only 863 of 866 completed. mysql-test-run: *** ERROR: Not all tests completed *** Error code 1 make: Fatal error: Command failed for target `test-ns'

2 1

[Maria-developers] order by x limit y optimization
by Tonu Samuel 26 Apr '09

26 Apr '09

Hi! Just was looking at processlist and thinking "why he cannot optimize such a easy case". I think there is a possible BOOST place (this is common query in many apps): mysql> select count(*) from url; +----------+ | count(*) | +----------+ | 17945916 | +----------+ 1 row in set (0.00 sec) mysql> mysql> show full processlist; +------+------+-----------+-----------+---------+------+----------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | Id | User | Host | db | Command | Time | State | Info | +------+------+-----------+-----------+---------+------+----------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | 78 | root | localhost | eeindexer | Query | 141 | Sorting result | INSERT INTO udm_url_tmp SELECT url.rec_id FROM url WHERE next_index_time<=1240742878 ORDER BY hops LIMIT 256 ..... So, it takes HUGE time to find out 256 rows with least hops. How I would make it manually (MySQL engine has internally better ways to do it): mysql> select min(hops) from url; +-----------+ | min(hops) | +-----------+ | 0 | +-----------+ 1 row in set (0.12 sec) mysql> select count(hops) from url where hops=0; +-------------+ | count(hops) | +-------------+ | 1 | +-------------+ 1 row in set (0.00 sec) mysql> select count(hops) from url where hops=1; +-------------+ | count(hops) | +-------------+ | 92 | +-------------+ 1 row in set (0.00 sec) mysql> select count(hops) from url where hops=2; +-------------+ | count(hops) | +-------------+ | 9790 | +-------------+ 1 row in set (0.00 sec) mysql> mysql> SELECT url.rec_id FROM url WHERE hops in (0,1,2,3) ORDER BY hops LIMIT 256; +--------+ | rec_id | +--------+ | 1 | | 82 | | 121 | | 55 | | 63 | | 39 | | 65 | | 100 | | 129 | | 85 | | 25 | | 11 | | 12 | | 54 | | 3 | | 101 | | 7 | | 50 | | 76 | | 77 | | 10 | | 24 | | 14 | | 2 | | 57 | | 46 | | 44 | | 17 | | 122 | | 135 | | 8 | | 51 | | 104 | | 124 | | 35 | | 56 | | 116 | | 62 | | 41 | | 78 | | 126 | | 15 | | 26 | | 40 | | 90 | | 80 | | 13 | | 36 | | 45 | | 70 | | 60 | | 117 | | 47 | | 114 | | 132 | | 19 | | 96 | | 118 | | 74 | | 9 | | 120 | | 95 | | 23 | | 136 | | 71 | | 130 | | 83 | | 33 | | 111 | | 125 | | 81 | | 43 | | 97 | | 28 | | 73 | | 109 | | 98 | | 105 | | 133 | | 128 | | 106 | | 4 | | 102 | | 127 | | 5 | | 30 | | 115 | | 123 | | 27 | | 6 | | 113 | | 93 | | 48 | | 10047 | | 55273 | | 7120 | | 21448 | | 16872 | | 17985 | | 9862 | | 20979 | | 45717 | | 491 | | 16943 | | 18183 | | 1753 | | 21176 | | 16486 | | 7578 | | 2329 | | 9824 | | 1239 | | 400 | | 6470 | | 8052 | | 2752 | | 1258 | | 16039 | | 593 | | 49887 | | 5884 | | 7119 | | 3011 | | 7819 | | 14710 | | 244 | | 6881 | | 16518 | | 240 | | 16445 | | 49988 | | 62181 | | 61227 | | 61651 | | 18118 | | 9951 | | 54964 | | 61197 | | 16361 | | 21452 | | 15361 | | 2840 | | 16479 | | 17006 | | 2768 | | 7917 | | 61198 | | 21556 | | 2650 | | 2525 | | 55022 | | 54911 | | 55672 | | 15326 | | 14763 | | 15493 | | 55156 | | 21435 | | 49959 | | 18242 | | 1887 | | 6437 | | 19407 | | 7593 | | 2355 | | 16908 | | 16102 | | 1275 | | 57868 | | 15943 | | 7977 | | 16079 | | 14712 | | 49874 | | 17269 | | 55851 | | 57783 | | 3442 | | 18085 | | 50913 | | 4067 | | 19402 | | 2785 | | 7920 | | 16456 | | 19143 | | 15141 | | 54895 | | 3578 | | 2905 | | 16623 | | 16802 | | 6472 | | 5894 | | 7004 | | 14711 | | 2545 | | 19184 | | 2994 | | 3297 | | 7993 | | 10034 | | 51738 | | 16854 | | 7007 | | 7583 | | 15967 | | 3720 | | 54940 | | 15299 | | 2343 | | 9896 | | 7016 | | 19668 | | 16128 | | 21624 | | 7529 | | 3521 | | 16602 | | 16089 | | 54821 | | 51589 | | 61828 | | 19690 | | 3608 | | 17946 | | 61146 | | 2265 | | 9809 | | 2509 | | 54729 | | 2426 | | 21617 | | 7543 | | 1841 | | 3700 | | 21391 | | 21442 | | 55155 | | 7532 | | 3490 | | 7821 | | 1891 | | 9876 | | 3677 | | 10019 | | 2862 | | 16186 | | 16299 | | 16528 | | 50641 | | 5926 | | 55135 | | 16800 | | 54734 | | 19375 | +--------+ 256 rows in set (0.09 sec) mysql> So why not to look on index cardinality, limit size (let say, less than 1% of table), min or max value of field ordered by and then exclude big part of table to avoid unnecessary work? Tõnu

1 0

[Maria-developers] Failures in mysql_client_test
by Kristian Nielsen 24 Apr '09

24 Apr '09

Hi Igor, I have been trying to figure out the problems you see in the mysql_client_test test case in the MariaDB 5.1 tree. It has been tricky due to not being able to reproduce. Patrick Crews was able to reproduce though, and was helpful enough to file this bug with some additional info: https://bugs.launchpad.net/maria/+bug/363181 >From this information, I pushed a few minor fixes to the Launchpad tree (two missing DBUG_RETURN statements). After this push, Patrick can no longer reproduce the failure. So, while my fixes were done in the area of the code where the problem seems to occur, I don't understand why two missing DBUG_RETURN would cause the error we saw... Maybe you can try again to see if you can reproduce the bug in the latest tree? If you cannot, then we may have to consider the problem fixed for now. If you can still reproduce, it would be very helpful if you could reproduce it with dbug enabled: perl mysql-test-run.pl --debug mysql_client_test and collect all the *.trace and *.err files from mysql-test/var/ and attach them to the bug or email them. It would be good to get progress on this annoying problem. - Kristian.

1 0

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2697)
by knielsen＠knielsen-hq.org 23 Apr '09

23 Apr '09

#At lp:maria 2697 knielsen(a)knielsen-hq.org 2009-04-23 Bump maria version following release. modified: configure.in per-file messages: configure.in Bump maria version following release. === modified file 'configure.in' --- a/configure.in 2009-03-22 12:16:09 +0000 +++ b/configure.in 2009-04-23 13:06:16 +0000 @@ -10,7 +10,7 @@ AC_CANONICAL_SYSTEM # # When changing major version number please also check switch statement # in mysqlbinlog::check_master_version(). -AM_INIT_AUTOMAKE(mysql, 5.1.32-maria-beta) +AM_INIT_AUTOMAKE(mysql, 5.1.32-maria-beta2) AM_CONFIG_HEADER([include/config.h:config.h.in]) PROTOCOL_VERSION=10

1 0

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2696)
by knielsen＠knielsen-hq.org 21 Apr '09

21 Apr '09

#At lp:maria 2696 knielsen(a)knielsen-hq.org 2009-04-21 [merge] merged from trunk. modified: sql/mysqld.cc sql/sql_string.cc sql/strfunc.cc === modified file 'sql/mysqld.cc' --- a/sql/mysqld.cc 2009-04-08 16:55:26 +0000 +++ b/sql/mysqld.cc 2009-04-08 21:16:10 +0000 @@ -4975,11 +4975,11 @@ inline void kill_broken_server() #if !defined(__NETWARE__) unix_sock == INVALID_SOCKET || #endif - (!opt_disable_networking && - (base_ip_sock == INVALID_SOCKET || extra_ip_sock != INVALID_SOCKET))) + (!opt_disable_networking && base_ip_sock == INVALID_SOCKET)) { select_thread_in_use = 0; /* The following call will never return */ + DBUG_PRINT("general", ("killing server because socket is closed")); kill_server(IF_NETWARE(MYSQL_KILL_SIGNAL, (void*) MYSQL_KILL_SIGNAL)); } } === modified file 'sql/sql_string.cc' --- a/sql/sql_string.cc 2008-02-29 13:56:50 +0000 +++ b/sql/sql_string.cc 2009-04-08 20:08:38 +0000 @@ -65,10 +65,10 @@ bool String::real_alloc(uint32 arg_lengt bool String::realloc(uint32 alloc_length) { - uint32 len=ALIGN_SIZE(alloc_length+1); - if (Alloced_length < len) + if (Alloced_length <= alloc_length) { char *new_ptr; + uint32 len= ALIGN_SIZE(alloc_length+1); if (alloced) { if ((new_ptr= (char*) my_realloc(Ptr,len,MYF(MY_WME)))) === modified file 'sql/strfunc.cc' --- a/sql/strfunc.cc 2007-08-13 13:11:25 +0000 +++ b/sql/strfunc.cc 2009-04-08 20:08:38 +0000 @@ -38,7 +38,8 @@ static const char field_separator=','; -ulonglong find_set(TYPELIB *lib, const char *str, uint length, CHARSET_INFO *cs, +ulonglong find_set(TYPELIB *lib, const char *str, uint length, + CHARSET_INFO *cs, char **err_pos, uint *err_len, bool *set_warning) { CHARSET_INFO *strip= cs ? cs : &my_charset_latin1;

1 0

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2695)
by knielsen＠knielsen-hq.org 21 Apr '09

21 Apr '09

#At lp:maria 2695 knielsen(a)knielsen-hq.org 2009-04-21 Fix two missing DBUG_RETURN. modified: libmysql/libmysql.c tests/mysql_client_test.c === modified file 'libmysql/libmysql.c' --- a/libmysql/libmysql.c 2008-12-10 09:02:25 +0000 +++ b/libmysql/libmysql.c 2009-04-21 10:24:51 +0000 @@ -4634,7 +4634,7 @@ int STDCALL mysql_stmt_fetch_column(MYSQ if ((int) stmt->state < (int) MYSQL_STMT_FETCH_DONE) { set_stmt_error(stmt, CR_NO_DATA, unknown_sqlstate, NULL); - return 1; + DBUG_RETURN(1); } if (column >= stmt->field_count) { === modified file 'tests/mysql_client_test.c' --- a/tests/mysql_client_test.c 2009-02-19 09:01:25 +0000 +++ b/tests/mysql_client_test.c 2009-04-21 10:24:51 +0000 @@ -16295,6 +16295,8 @@ static void test_bug27876() rc= mysql_query(mysql, "set names default"); myquery(rc); + + DBUG_VOID_RETURN; }

1 0

[Maria-developers] MariaDB source and binaries tarballs made available
by Kristian Nielsen 20 Apr '09

20 Apr '09

As the MySQL Users's Conference is starting up, Monty and I wanted to have tarballs with source and binaries available to make it easier for interested people to try it out. The plan for future binaries is to have Ourdelta publish them from MariaDB sources, however due to lack of time this was not possible to set up before the conference. So to at least have something, I have built a source tarball from the latest bzr repository of MariaDB 5.1, as well as generic binary packages for 32- and 64-bit x86 Linux. I made the tarballs available here: http://askmonty.org/wiki/index.php/MariaDB:Download I took some care to get the build options correct and to check that everything works ok. However, I cannot check everything; notably I do not have access to anything but 32- and 64-bit x86 Ubuntu 8.04 systems to test on. I checked that the test suite passes, that the major storage engines work (MyISAM, Maria, InnoDB, PBXT), that the major features are there (partitioning, ssl, pool-of-threads), and that the binary tarballs can be installed and run. Any additional help with testing the packages for possible problems with be much appreciated, of course. The packages are currently hosted on a 5Mbit uplink I believe, so I will next work to get mirroring set up with the help of those in the community that have already offered to help with this. - Kristian.

1 0

Re: [Maria-developers] Maria binary builds
by Kristian Nielsen 19 Apr '09

19 Apr '09

Michael Widenius <michael.widenius(a)gmail.com> writes: > Hi! > >>>>>> "Kristian" == Kristian Nielsen <knielsen(a)knielsen-hq.org> writes: > > <cut> > > Kristian> I am still not sure whether generic (.tar.gz) binaries should be build > Kristian> statically (and to what extend). > > The benefit of building them staticly are: Unfortunately, the issue is more complex than just this. A fully static binary will not work at all in a number of scenarios. A fully static binary is obtained by linking everything statically, including libc. However, linking libc statically creates a number of problems if mysqld tries to load additional .so files when it is run. The additional .so files then use a different version of libc than the one mysqld code uses, which does not work. This gives problem loading plugins (like UDFs), and on many systems causes problems even without loading any plugins, as libc loads .so files itself (notably for NSS, domain name resolution). There are a number of bugs on this on bugs.mysql.com. Currently, I'm linking libc dynamically to avoid these problems (I believe/hope that glibc binary compatibility has become fairly good by now). I'm linking libz statically (bundled version) as well as libgcc as this shouldn't cause similar problems. And compiling with `gcc` instead of `g++` to avoid linking libstc++ at all. All mariadb code is linked statically as well, of course. I checked, and this seems to be the same that is done in the Sun/MySQL generic binaries these days. - Kristian.

1 0