- developers - lists.mariadb.org

Re: [Maria-developers] [Commits] 27af988: MDEV-11479 Improved wsrep_dirty_reads
by Vicențiu Ciorbaru 13 Dec '16

13 Dec '16

Hi Sachin! Comments inline: On Sun, 4 Dec 2016 at 08:36 <sachin.setiya(a)mariadb.com> wrote: > revision-id: 27af98805c27c8e0c41dc8c19bf80aefff4c6d3d > (mariadb-galera-5.5.53-3-g27af988) > parent(s): 72fd15f7c31aa3e3705ae1b005a3247a985c5bb3 > author: SachinSetiya > committer: SachinSetiya > NIT: To get these to display correctly you should have a space between Sachin and Setiya. > timestamp: 2016-12-04 12:05:05 +0530 > message: > > MDEV-11479 Improved wsrep_dirty_reads > > Tasks:- > Changes in wsrep_dirty_reads variable > 1.) Global + Session scope (Current: session-only) > 2.) Allow all commands that do not change data (besides SELECT) > 3.) Allow prepared Statements that do not change data > 4.) Works with wsrep_sync_wait enabled > You also seemed to have made the variable available as a command line option. Please specify that as well. > --- > .../suite/galera/r/galera_var_dirty_reads.result | 29 > +++++++++++++++++++--- > .../suite/galera/t/galera_var_dirty_reads.test | 25 > ++++++++++++++++++- > .../sys_vars/r/wsrep_dirty_reads_basic.result | 19 ++++++++++++-- > .../suite/sys_vars/t/wsrep_dirty_reads_basic.test | 13 ++++++++-- > sql/sql_parse.cc | 18 ++++++++++---- > sql/sys_vars.cc | 4 +-- > sql/wsrep_mysqld.cc | 3 ++- > sql/wsrep_mysqld.h | 1 + > 8 files changed, 96 insertions(+), 16 deletions(-) > > diff --git a/mysql-test/suite/galera/r/galera_var_dirty_reads.result > b/mysql-test/suite/galera/r/galera_var_dirty_reads.result > index 6a2aa1e..d009c1b 100644 > --- a/mysql-test/suite/galera/r/galera_var_dirty_reads.result > +++ b/mysql-test/suite/galera/r/galera_var_dirty_reads.result > @@ -28,9 +28,32 @@ SELECT @@max_allowed_packet; > SELECT 2+2 from DUAL; > 2+2 > 4 > -SELECT sysdate() from DUAL; > -sysdate() > -2016-10-28 23:13:06 > +SET @@session.wsrep_dirty_reads=ON; > +SELECT * FROM t1; > +i > +1 > +prepare stmt_show from 'select 1'; > +prepare stmt_select from 'select * from t1'; > +prepare stmt_insert from 'insert into t1 values(1)'; > +execute stmt_show; > +1 > +1 > +execute stmt_select; > +i > +1 > +execute stmt_insert; > +ERROR 08S01: WSREP has not yet prepared node for application use > +select * from t1; > +i > +1 > +execute stmt_show; > +1 > +1 > +execute stmt_select; > +i > +1 > +execute stmt_insert; > +ERROR 08S01: WSREP has not yet prepared node for application use > SELECT * FROM t1; > i > 1 > You're not really testing the global aspect of the variable here. You're only testing the sesssion behaviour of it. Also, test with it both ON and OFF (when setting it globally). Here's a sample that tests both session and global behaviour: --enable_connect_log SELECT @@expensive_subquery_limit; SET @@session.expensive_subquery_limit=300; SELECT @@expensive_subquery_limit; create user user1; create user user2; --connect (con1, localhost, user1,,) SELECT @@expensive_subquery_limit; SET @@session.expensive_subquery_limit=300; SELECT @@expensive_subquery_limit; connection default; SELECT @@expensive_subquery_limit; SET @@global.expensive_subquery_limit=400; SELECT @@expensive_subquery_limit; --connect (con2, localhost, user1,,) SELECT @@expensive_subquery_limit; connection default; drop user user1; drop user user2; ************************************ SELECT @@expensive_subquery_limit; @@expensive_subquery_limit 100 SET @@session.expensive_subquery_limit=300; SELECT @@expensive_subquery_limit; @@expensive_subquery_limit 300 create user user1; create user user2; connect con1, localhost, user1,,; SELECT @@expensive_subquery_limit; @@expensive_subquery_limit 100 SET @@session.expensive_subquery_limit=300; SELECT @@expensive_subquery_limit; @@expensive_subquery_limit 300 connection default; SELECT @@expensive_subquery_limit; @@expensive_subquery_limit 300 SET @@global.expensive_subquery_limit=400; SELECT @@expensive_subquery_limit; @@expensive_subquery_limit 300 connect con2, localhost, user1,,; SELECT @@expensive_subquery_limit; @@expensive_subquery_limit 400 connection default; drop user user1; drop user user2; Notice how the last connection has the default value 400 as that's the global value. For your use case, you'd also have to execute the prepared statements to see if it works (execute select_stmt). > diff --git a/mysql-test/suite/galera/t/galera_var_dirty_reads.test > b/mysql-test/suite/galera/t/galera_var_dirty_reads.test > index dd7bc88..44931aa 100644 > --- a/mysql-test/suite/galera/t/galera_var_dirty_reads.test > +++ b/mysql-test/suite/galera/t/galera_var_dirty_reads.test > @@ -44,7 +44,30 @@ SET @@session.wsrep_dirty_reads=OFF; > SELECT 2; > SELECT @@max_allowed_packet; > SELECT 2+2 from DUAL; > -SELECT sysdate() from DUAL; > + > +#Query which does not change data should be allowed MDEV-11479 > +SET @@session.wsrep_dirty_reads=ON; > +SELECT * FROM t1; > + > +#Prepared statement creation should be allowed MDEV-11479 > +prepare stmt_show from 'select 1'; > +prepare stmt_select from 'select * from t1'; > +prepare stmt_insert from 'insert into t1 values(1)'; > + > +#Only prepare statement which does not change data should be allowed > +execute stmt_show; > +execute stmt_select; > +--error ER_UNKNOWN_COM_ERROR > +execute stmt_insert; + > +#wsrep_dirty_read should work when wsrep_sync_wait is 1 or non zero > +#because we already are disconnected , So It does not make any sense > +#to wait for other nodes > I don't think you're testing this correctly. See my comment on the if statement which is supposed to govern this. > +select * from t1; > +execute stmt_show; > +execute stmt_select; > +--error ER_UNKNOWN_COM_ERROR > +execute stmt_insert; > > --disable_query_log > --eval SET @@global.wsrep_cluster_address = '$wsrep_cluster_address_saved' > diff --git a/mysql-test/suite/sys_vars/r/wsrep_dirty_reads_basic.result > b/mysql-test/suite/sys_vars/r/wsrep_dirty_reads_basic.result > index d2a62d6..649ac83 100644 > --- a/mysql-test/suite/sys_vars/r/wsrep_dirty_reads_basic.result > +++ b/mysql-test/suite/sys_vars/r/wsrep_dirty_reads_basic.result > @@ -5,12 +5,13 @@ > SET @wsrep_dirty_reads_session_saved = @@session.wsrep_dirty_reads; > # default > SELECT @@global.wsrep_dirty_reads; > -ERROR HY000: Variable 'wsrep_dirty_reads' is a SESSION variable > +@@global.wsrep_dirty_reads > +0 > SELECT @@session.wsrep_dirty_reads; > @@session.wsrep_dirty_reads > 0 > > -# scope and valid values > +# valid values for session > SET @@session.wsrep_dirty_reads=OFF; > SELECT @@session.wsrep_dirty_reads; > @@session.wsrep_dirty_reads > @@ -24,6 +25,20 @@ SELECT @@session.wsrep_dirty_reads; > @@session.wsrep_dirty_reads > 0 > > +# valid values for global > +SET @@global.wsrep_dirty_reads=OFF; > +SELECT @@global.wsrep_dirty_reads; > +@@global.wsrep_dirty_reads > +0 > +SET @@global.wsrep_dirty_reads=ON; > +SELECT @@global.wsrep_dirty_reads; > +@@global.wsrep_dirty_reads > +1 > +SET @@global.wsrep_dirty_reads=default; > +SELECT @@global.wsrep_dirty_reads; > +@@global.wsrep_dirty_reads > +0 > + > # invalid values > SET @@session.wsrep_dirty_reads=NULL; > ERROR 42000: Variable 'wsrep_dirty_reads' can't be set to the value of > 'NULL' > Test invalid values for global scope as well. It should not change obviously but since we're writing tests, might as well do that too. > diff --git a/mysql-test/suite/sys_vars/t/wsrep_dirty_reads_basic.test > b/mysql-test/suite/sys_vars/t/wsrep_dirty_reads_basic.test > index a47524f..000c4c4 100644 > --- a/mysql-test/suite/sys_vars/t/wsrep_dirty_reads_basic.test > +++ b/mysql-test/suite/sys_vars/t/wsrep_dirty_reads_basic.test > @@ -8,12 +8,12 @@ > SET @wsrep_dirty_reads_session_saved = @@session.wsrep_dirty_reads; > > --echo # default > ---error ER_INCORRECT_GLOBAL_LOCAL_VAR > + > SELECT @@global.wsrep_dirty_reads; > SELECT @@session.wsrep_dirty_reads; > > --echo > ---echo # scope and valid values > +--echo # valid values for session > SET @@session.wsrep_dirty_reads=OFF; > SELECT @@session.wsrep_dirty_reads; > SET @@session.wsrep_dirty_reads=ON; > @@ -22,6 +22,15 @@ SET @@session.wsrep_dirty_reads=default; > SELECT @@session.wsrep_dirty_reads; > > --echo > +--echo # valid values for global > +SET @@global.wsrep_dirty_reads=OFF; > +SELECT @@global.wsrep_dirty_reads; > +SET @@global.wsrep_dirty_reads=ON; > +SELECT @@global.wsrep_dirty_reads; > +SET @@global.wsrep_dirty_reads=default; > +SELECT @@global.wsrep_dirty_reads; > + > +--echo > --echo # invalid values > --error ER_WRONG_VALUE_FOR_VAR > SET @@session.wsrep_dirty_reads=NULL; > diff --git a/sql/sql_parse.cc b/sql/sql_parse.cc > index eb4f714..a56d694 100644 > --- a/sql/sql_parse.cc > +++ b/sql/sql_parse.cc > @@ -2393,10 +2393,11 @@ mysql_execute_command(THD *thd) > */ > if (thd->variables.wsrep_on && !thd->wsrep_applier && !wsrep_ready && > lex->sql_command != SQLCOM_SET_OPTION && > - !(thd->variables.wsrep_dirty_reads && > - lex->sql_command == SQLCOM_SELECT) && > - !(lex->sql_command == SQLCOM_SELECT && > - !all_tables) && > + !(thd->variables.wsrep_dirty_reads && > + (sql_command_flags[lex->sql_command] & > + CF_CHANGES_DATA) == 0) && > Put this line on one line only, the single & is confusing, or better, turn it into a small inline function, such as command_changes_data() > + !(lex->sql_command == SQLCOM_SELECT && > + !all_tables) && > Can't tell for certain but this condition might not be necessary if you have the previous one. What happens when you have a stored procedure that inserts into a table? Does the CF_CHANGES_DATA flag filter out the query or not? ex: select function_with_side_efect(1); This seems like a good test case that I'm not sure is tested for this variable anyway. Please try it. > !wsrep_is_show_query(lex->sql_command)) > { > #if DIRTY_HACK > @@ -2512,7 +2513,14 @@ mysql_execute_command(THD *thd) > case SQLCOM_SHOW_INDEX_STATS: > case SQLCOM_SELECT: > #ifdef WITH_WSREP > - if (WSREP_CLIENT(thd) && wsrep_sync_wait(thd)) goto error; > + /* > + wsrep_dirty_read should work when wsrep_sync_wait is 1 or non-zero > + because we already are disconnected. So, it does not make any sense > + to wait for other nodes > + */ > > This here is not correct. This will not execute the select if wsrep_dirty_reads is true and wsrep_ready is true and wsrep_sync_wait(thd) is non-zero. if (WSREP_CLIENT(thd) && thd->variables.wsrep_dirty_reads(wsrep_ready > + if (WSREP_CLIENT(thd) && (wsrep_ready || > !thd->variables.wsrep_dirty_reads) > + && wsrep_sync_wait(thd)) > + goto error case SQLCOM_SHOW_VARIABLES: > case SQLCOM_SHOW_CHARSETS: > case SQLCOM_SHOW_COLLATIONS: > diff --git a/sql/sys_vars.cc b/sql/sys_vars.cc > index bdcedff..0d864f4 100644 > --- a/sql/sys_vars.cc > +++ b/sql/sys_vars.cc > @@ -3967,8 +3967,8 @@ static Sys_var_mybool Sys_wsrep_restart_slave( > > static Sys_var_mybool Sys_wsrep_dirty_reads( > "wsrep_dirty_reads", "Do not reject SELECT queries even when the > node " > - "is not ready.", SESSION_ONLY(wsrep_dirty_reads), NO_CMD_LINE, > - DEFAULT(FALSE), NO_MUTEX_GUARD, NOT_IN_BINLOG); > + "is not ready.", SESSION_VAR(wsrep_dirty_reads), + CMD_LINE(OPT_ARG), DEFAULT(FALSE), NO_MUTEX_GUARD, NOT_IN_BINLOG); > We might need a guard here since global variables can be changed from multiple places. Not 100% sure about this but that's my intuition. Need to look this up. > #endif /* WITH_WSREP */ > > diff --git a/sql/wsrep_mysqld.cc b/sql/wsrep_mysqld.cc > index 48114fc..f21ff92 100644 > --- a/sql/wsrep_mysqld.cc > +++ b/sql/wsrep_mysqld.cc > @@ -65,7 +65,8 @@ my_bool wsrep_restart_slave_activated = 0; // node has > dropped, and slave > // restart will be needed > my_bool wsrep_slave_UK_checks = 0; // slave thread does UK checks > my_bool wsrep_slave_FK_checks = 0; // slave thread does FK checks > - > +bool wsrep_dirty_reads = 0; // is dirty reads allowed for > + // disconnectd/non major node > Typo: disconnectd -> disconnected. I really hate how the whole file is set up with TRUE/FALSE/0 for either bool or my_bool. It's so inconsistent. I'd prefer 'true' here to complicate stuff further. :) To be "consistent" I'd make it a my_bool, but we should clean up all of this mess at one point. > /* > Set during the creation of first wsrep applier and rollback threads. > Since these threads are critical, abort if the thread creation fails. > diff --git a/sql/wsrep_mysqld.h b/sql/wsrep_mysqld.h > index 3fb5189..9b9179a 100644 > --- a/sql/wsrep_mysqld.h > +++ b/sql/wsrep_mysqld.h > @@ -85,6 +85,7 @@ extern ulong wsrep_max_ws_size; > extern ulong wsrep_max_ws_rows; > extern const char* wsrep_notify_cmd; > extern my_bool wsrep_certify_nonPK; > +extern bool wsrep_dirty_reads; > Remember to update the type here too. > extern long wsrep_max_protocol_version; > extern long wsrep_protocol_version; > extern ulong wsrep_forced_binlog_format;

2 1

Re: [Maria-developers] [Commits] 514b557: MDEV-11170: MariaDB 10.2 cannot start on MySQL 5.7 datadir:
by Sergei Golubchik 12 Dec '16

12 Dec '16

Hi, Vicentiu! I think mariadb should support 5.7 user table (without the password column) without any errors. That shouldn't be too difficult, just detect at run-time whether the column "Password" exists and adjust all other column numbers accordingly. On Dec 12, Vicentiu wrote: > revision-id: 514b557d066fcd1a90ccbbcdbaf5fa5fe65311eb (mariadb-10.2.2-138-g514b557) > parent(s): 9320d8ae30c18420bef659618175836221d363ea > author: Vicențiu Ciorbaru > committer: Vicențiu Ciorbaru > timestamp: 2016-12-12 04:26:27 +0200 > message: > > MDEV-11170: MariaDB 10.2 cannot start on MySQL 5.7 datadir: > > Fatal error: mysql.user table is damaged or in unsupported 3.20 format > > The problem stems from MySQL 5.7.6. According to MySQL documentation: > In MySQL 5.7.6, the Password column was removed and all credentials are > stored in the authentication_string column. > > If opening a MySQL 5.7.6 (and up) datadir with MariaDB 10.2, the user table > appears corrupted. In order to fix this, the server must be started with > --skip-grant-tables and then a subsequent mysql_upgrade command must be > issued. > > This patch updates the mysql_upgrade command to also add the removed > Password column. The password column is necessary, otherwise > the mysql_upgrade script fails due to the Event_scheduler not being able > to start, as it can't find Event_priv in the table where it ought to be. > MySQL's version has column position 28 (0 index) vs our datadir version > expects position 29. > > --- > scripts/mysql_system_tables_fix.sql | 6 ++++++ > 1 file changed, 6 insertions(+) > Regards, Sergei Chief Architect MariaDB and security(a)mariadb.org

1 0

Re: [Maria-developers] 02e74a5: MDEV-11700: MariaDB 10.2 cannot start on MySQL 5.7 datadir:
by Vicențiu Ciorbaru 12 Dec '16

12 Dec '16

Hi Sergei! Can you please review this patch for MDEV-11700? Details presented in the commit message. I'm not sure which solution we want. Possible solutions: A) have the user fix it by first starting the server with --skip-grant-tables to avoid "mysql.user table is damaged", then run mysql_upgrade, to fix the problem. Without this patch, this solution fails at mysql_upgrade. B) Have the server understand MySQL 5.7.6 + table format where the Password column is missing. For 10.2 I've chosen solution A. Although this should probably also be targeted for 10.1? For 10.3, since we're planning on doing some authentication changes, we can implement a variant of B. Thoughts? On Mon, 12 Dec 2016 at 03:50 vicentiu <vicentiu(a)mariadb.org> wrote: > revision-id: 02e74a5614b08cfce505002abc26d4634648308e > (mariadb-10.2.2-138-g02e74a5) > parent(s): 9320d8ae30c18420bef659618175836221d363ea > author: Vicențiu Ciorbaru > committer: Vicențiu Ciorbaru > timestamp: 2016-12-12 03:42:11 +0200 > message: > > MDEV-11700: MariaDB 10.2 cannot start on MySQL 5.7 datadir: > > Fatal error: mysql.user table is damaged or in unsupported 3.20 format > > The problem stems from MySQL 5.7.6. According to MySQL documentation: > In MySQL 5.7.6, the Password column was removed and all credentials are > stored in the authentication_string column. > > If opening a MySQL 5.7.6 (and up) datadir with MariaDB 10.2, the user table > appears corrupted. In order to fix this, the server must be started with > --skip-grant-tables and then a subsequent mysql_upgrade command must be > issued. > > This patch updates the mysql_upgrade command to also add the removed > Password column. The password column is necessary, otherwise > the mysql_upgrade script fails due to the Event_scheduler not being able > to start, as it can't find Event_priv in the table where it ought to be. > MySQL's version has column position 28 (0 index) vs our datadir version > expects position 29. > > --- > scripts/mysql_system_tables_fix.sql | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/scripts/mysql_system_tables_fix.sql > b/scripts/mysql_system_tables_fix.sql > index ea1059c..80d0ef7 100644 > --- a/scripts/mysql_system_tables_fix.sql > +++ b/scripts/mysql_system_tables_fix.sql > @@ -164,6 +164,12 @@ ALTER TABLE user > MODIFY Host char(60) NOT NULL default '', > MODIFY User char(80) NOT NULL default '', > ENGINE=MyISAM, CONVERT TO CHARACTER SET utf8 COLLATE utf8_bin; > + > +# In MySQL 5.7.6 the Password column is removed. Recreate it to preserve > the number > +# of columns MariaDB expects in the user table. > +ALTER TABLE user > + ADD Password char(41) character set latin1 collate latin1_bin NOT NULL > default '' AFTER User; > + > ALTER TABLE user > MODIFY Password char(41) character set latin1 collate latin1_bin NOT > NULL default '', > MODIFY Select_priv enum('N','Y') COLLATE utf8_general_ci DEFAULT 'N' > NOT NULL, >

2 2

[Maria-developers] Review for version#2 of MDEV-11297: Add support for LIMIT clause in GROUP_CONCAT
by Sergey Petrunia 11 Dec '16

11 Dec '16

Hi Varun, Nice to see that nearly all comments from the previous review were addressed. Please find my further input below. General questions: How is DISTINCT handled? Syntax like GROUP_CONCAT(DISTINCT col ORDER BY ... LIMIT ...) is allowed by the parser, however I don't see any handling for DISTINCT in your patch. Did you forget to update group_concat_limit.result file? Please check. > diff -urpN '--exclude=.*' 10.2-varun-mdev11297-orig/sql/item_sum.cc 10.2-varun-mdev11297-noc/sql/item_sum.cc > --- 10.2-varun-mdev11297-orig/sql/item_sum.cc 2016-12-11 12:17:04.023961449 +0300 > +++ 10.2-varun-mdev11297-noc/sql/item_sum.cc 2016-12-11 12:18:40.411957365 +0300 > @@ -3114,6 +3114,11 @@ int dump_leaf_key(void* key_arg, element > String *result= &item->result; > Item **arg= item->args, **arg_end= item->args + item->arg_count_field; > uint old_length= result->length(); > + ulonglong *offset_limit= &item->copy_offset_limit; > + ulonglong *row_limit = &item->copy_row_limit; > + > + if (item->limit_clause && !(*row_limit)) > + return 1; > > if (item->no_appended) > item->no_appended= FALSE; > @@ -3122,20 +3127,23 @@ int dump_leaf_key(void* key_arg, element > > tmp.length(0); > > - for (; arg < arg_end; arg++) > + if (!item->limit_clause || !(*offset_limit)) > { > - String *res; > - /* > - We have to use get_tmp_table_field() instead of > - real_item()->get_tmp_table_field() because we want the field in > - the temporary table, not the original field > - We also can't use table->field array to access the fields > - because it contains both order and arg list fields. > - */ > - if ((*arg)->const_item()) > - res= (*arg)->val_str(&tmp); > - else > + > + for (; arg < arg_end; arg++) > { > + String *res; > + /* > + We have to use get_tmp_table_field() instead of > + real_item()->get_tmp_table_field() because we want the field in > + the temporary table, not the original field > + We also can't use table->field array to access the fields > + because it contains both order and arg list fields. > + */ > + if ((*arg)->const_item()) > + res= (*arg)->val_str(&tmp); > + else > + { > Field *field= (*arg)->get_tmp_table_field(); Please make the code to use the right indentation. > if (field) > { > @@ -3146,9 +3154,19 @@ int dump_leaf_key(void* key_arg, element > } > else > res= (*arg)->val_str(&tmp); > + } > + if (res) > + { > + result->append(*res); > + } > } > - if (res) > - result->append(*res); > + if ((*row_limit)) > + (*row_limit)--; > + } > + else > + { > + item->no_appended= TRUE; If I comment out the above line, the test still passes. Please add test coverage that shows it is needed, or remove it. > + (*offset_limit)--; > } > > item->row_count++; > @@ -3199,7 +3217,8 @@ Item_func_group_concat:: > Item_func_group_concat(THD *thd, Name_resolution_context *context_arg, > bool distinct_arg, List<Item> *select_list, > const SQL_I_List<ORDER> &order_list, > - String *separator_arg) > + String *separator_arg, bool limit_clause, > + Item *rowLimit, Item *offsetLimit) camelCase or javaStyleVariableNames are against the coding style. Please use something like row_limit_arg if there is no better option. (In general, please try to make your new code fit in with the rest of the code. Did you see camelCase anywhere? If not, it's a sign not to use it). > :Item_sum(thd), tmp_table_param(0), separator(separator_arg), tree(0), > unique_filter(NULL), table(0), > order(0), context(context_arg), > @@ -3208,7 +3227,9 @@ Item_func_group_concat(THD *thd, Name_re > row_count(0), > distinct(distinct_arg), > warning_for_row(FALSE), > - force_copy_fields(0), original(0) > + force_copy_fields(0),row_limit(0), > + offset_limit(0),limit_clause(limit_clause), > + copy_offset_limit(0), copy_row_limit(0),original(0) Please use 'NULL' for pointers, not '0'. > { > Item *item_select; > Item **arg_ptr; > @@ -3250,6 +3271,16 @@ Item_func_group_concat(THD *thd, Name_re > /* orig_args is only used for print() */ > orig_args= (Item**) (order + arg_count_order); > memcpy(orig_args, args, sizeof(Item*) * arg_count); > + > + if (limit_clause) > + { > + if (!(row_limit=(Item*) thd->alloc(sizeof(Item)))) > + return; I'm not sure what the above is supposed to mean? Why allocate sizeof(Item) bytes? > + row_limit= rowLimit; ... especially when the next thing you do is the above assignment, so you lose the value that you've got from the thd->alloc() call? > + if (!(offset_limit= (Item*)thd->alloc(sizeof(Item)))) > + return; > + offset_limit= offsetLimit; The same question as above. > + } > } > > > @@ -3269,7 +3300,9 @@ Item_func_group_concat::Item_func_group_ > warning_for_row(item->warning_for_row), > always_null(item->always_null), > force_copy_fields(item->force_copy_fields), > - original(item) > + row_limit(item->row_limit), offset_limit(item->offset_limit), > + limit_clause(item->limit_clause),copy_offset_limit(item->copy_offset_limit), > + copy_row_limit(item->copy_row_limit), original(item->original) > { > quick_group= item->quick_group; > result.set_charset(collation.collation); > @@ -3382,6 +3415,10 @@ void Item_func_group_concat::clear() > null_value= TRUE; > warning_for_row= FALSE; > no_appended= TRUE; > + if (offset_limit) > + copy_offset_limit= offset_limit->val_int(); > + if (row_limit) > + copy_row_limit= row_limit->val_int(); > if (tree) > reset_tree(tree); > if (unique_filter) > diff -urpN '--exclude=.*' 10.2-varun-mdev11297-orig/sql/item_sum.h 10.2-varun-mdev11297-noc/sql/item_sum.h > --- 10.2-varun-mdev11297-orig/sql/item_sum.h 2016-12-11 12:17:04.043961448 +0300 > +++ 10.2-varun-mdev11297-noc/sql/item_sum.h 2016-12-11 12:18:40.415957365 +0300 > @@ -1583,6 +1583,12 @@ class Item_func_group_concat : public It > bool always_null; > bool force_copy_fields; > bool no_appended; > + Item *row_limit; > + Item *offset_limit; > + bool limit_clause; > + ulonglong copy_offset_limit; > + ulonglong copy_row_limit; > + In previous review I have requested: >> Please add (short) comments describing each new member variable. And you still haven't done that. > /* > Following is 0 normal object and pointer to original one for copy > (to correctly free resources) > @@ -1602,7 +1608,8 @@ protected: > public: > Item_func_group_concat(THD *thd, Name_resolution_context *context_arg, > bool is_distinct, List<Item> *is_select, > - const SQL_I_List<ORDER> &is_order, String *is_separator); > + const SQL_I_List<ORDER> &is_order, String *is_separator, > + bool limit_clause, Item *row_limit, Item *offset_limit); > > Item_func_group_concat(THD *thd, Item_func_group_concat *item); > ~Item_func_group_concat(); > diff -urpN '--exclude=.*' 10.2-varun-mdev11297-orig/sql/sql_yacc.yy 10.2-varun-mdev11297-noc/sql/sql_yacc.yy > --- 10.2-varun-mdev11297-orig/sql/sql_yacc.yy 2016-12-11 12:17:04.107961446 +0300 > +++ 10.2-varun-mdev11297-noc/sql/sql_yacc.yy 2016-12-11 12:18:40.887957345 +0300 > @@ -1786,7 +1786,7 @@ bool my_yyoverflow(short **a, YYSTYPE ** > %type <num> > order_dir lock_option > udf_type opt_local opt_no_write_to_binlog > - opt_temporary all_or_any opt_distinct > + opt_temporary all_or_any opt_distinct opt_glimit_clause > opt_ignore_leaves fulltext_options union_option > opt_not > select_derived_init transaction_access_mode_types > @@ -1844,7 +1844,7 @@ bool my_yyoverflow(short **a, YYSTYPE ** > opt_escape > sp_opt_default > simple_ident_nospvar simple_ident_q > - field_or_var limit_option > + field_or_var limit_option glimit_option > part_func_expr > window_func_expr > window_func > @@ -10449,16 +10449,34 @@ sum_expr: > | GROUP_CONCAT_SYM '(' opt_distinct > { Select->in_sum_expr++; } > expr_list opt_gorder_clause > - opt_gconcat_separator > + opt_gconcat_separator opt_glimit_clause > ')' > { > SELECT_LEX *sel= Select; > sel->in_sum_expr--; > - $$= new (thd->mem_root) > - Item_func_group_concat(thd, Lex->current_context(), $3, $5, > - sel->gorder_list, $7); > - if ($$ == NULL) > - MYSQL_YYABORT; > + if (!$8) > + $$= new (thd->mem_root) > + Item_func_group_concat(thd, Lex->current_context(), $3, $5, > + sel->gorder_list, $7, FALSE, 0, 0); > + else > + { > + if (sel->select_limit && sel->offset_limit) > + $$= new (thd->mem_root) > + Item_func_group_concat(thd, Lex->current_context(), $3, $5, > + sel->gorder_list, $7, TRUE, sel->select_limit, > + sel->offset_limit); > + else if (sel->select_limit) > + $$= new (thd->mem_root) > + Item_func_group_concat(thd, Lex->current_context(), $3, $5, > + sel->gorder_list, $7, TRUE, sel->select_limit, 0); > + } > + if ($$ == NULL) > + MYSQL_YYABORT; > + > + sel->select_limit= NULL; > + sel->offset_limit= NULL; > + sel->explicit_limit= 0; > + > $5->empty(); > sel->gorder_list.empty(); > } > @@ -10680,6 +10698,95 @@ gorder_list: > { if (add_gorder_to_list(thd, $1,(bool) $2)) MYSQL_YYABORT; } > ; What is the difference between glimit_option and limit_option? Looking at the grammar, they look the same. Please either remove glimit_option or explain why do you need it. I can see that limit_clause and glimit_clause have to be different because limit_clause also supports LIMIT ROWS EXAMINED. Actually, there is one thing in limit_clause that I think should be in glimit_clause also. It's this call: Lex->set_stmt_unsafe(LEX::BINLOG_STMT_UNSAFE_LIMIT); Did you omit it intentionally? Let's meet and discuss it. > +opt_glimit_clause: > + /* empty */ { $$ = 0; } > + | glimit_clause { $$ = 1; } > + ; > + > +glimit_clause_init: > + LIMIT{} > + ; > + > +glimit_clause: > + glimit_clause_init glimit_options{} > + ; > + > +glimit_options: > + glimit_option > + { > + SELECT_LEX *sel= Select; > + sel->select_limit= $1; > + sel->offset_limit= 0; > + sel->explicit_limit= 1; > + } > + | glimit_option ',' glimit_option > + { > + SELECT_LEX *sel= Select; > + sel->select_limit= $3; > + sel->offset_limit= $1; > + sel->explicit_limit= 1; > + } > + | glimit_option OFFSET_SYM glimit_option > + { > + SELECT_LEX *sel= Select; > + sel->select_limit= $1; > + sel->offset_limit= $3; > + sel->explicit_limit= 1; > + } > + ; > + > +glimit_option: > + ident > + { > + Item_splocal *splocal; > + LEX *lex= thd->lex; > + Lex_input_stream *lip= & thd->m_parser_state->m_lip; > + sp_variable *spv; > + sp_pcontext *spc = lex->spcont; > + if (spc && (spv = spc->find_variable($1, false))) > + { > + splocal= new (thd->mem_root) > + Item_splocal(thd, $1, spv->offset, spv->sql_type(), > + lip->get_tok_start() - lex->sphead->m_tmp_query, > + lip->get_ptr() - lip->get_tok_start()); > + if (splocal == NULL) > + MYSQL_YYABORT; > +#ifndef DBUG_OFF > + splocal->m_sp= lex->sphead; > +#endif > + lex->safe_to_cache_query=0; > + } > + else > + my_yyabort_error((ER_SP_UNDECLARED_VAR, MYF(0), $1.str)); > + if (splocal->type() != Item::INT_ITEM) > + my_yyabort_error((ER_WRONG_SPVAR_TYPE_IN_LIMIT, MYF(0))); > + splocal->limit_clause_param= TRUE; > + $$= splocal; > + } > + | param_marker > + { > + $1->limit_clause_param= TRUE; > + } > + | ULONGLONG_NUM > + { > + $$= new (thd->mem_root) Item_uint(thd, $1.str, $1.length); > + if ($$ == NULL) > + MYSQL_YYABORT; > + } > + | LONG_NUM > + { > + $$= new (thd->mem_root) Item_uint(thd, $1.str, $1.length); > + if ($$ == NULL) > + MYSQL_YYABORT; > + } > + | NUM > + { > + $$= new (thd->mem_root) Item_uint(thd, $1.str, $1.length); > + if ($$ == NULL) > + MYSQL_YYABORT; > + } > + ; > + > in_sum_expr: > opt_all > { BR Sergei -- Sergei Petrunia, Software Developer MariaDB Corporation | Skype: sergefp | Blog: http://s.petrunia.net/blog

1 0

[Maria-developers] Please review MDEV-11528 Split Item_func_min_max::val_xxx() and Item_func_min_max::get_date() into methods in Type_handler
by Alexander Barkov 10 Dec '16

10 Dec '16

Hello Alexey, Can you please review a patch for MDEV-11528. Thanks!

1 0

[Maria-developers] MDEV-11503 Introduce Type_handler::make_in_vector() and Item_func_in_fix_comparator_compatible_types()
by Alexander Barkov 09 Dec '16

09 Dec '16

Hello Sanja, Please review a patch for MDEV-11503. Thanks!

1 1

[Maria-developers] Fix for TokuDB and parallel replication
by Kristian Nielsen 09 Dec '16

09 Dec '16

Parallel replication so far did not work well with TokuDB, as some people who tried it found out. I have now pushed to 10.1 some patches to solve the problems. There are two main fixes: 1. Fix some races where a waiting transaction would miss its wakeup and get a lock timeout on a waiting row lock, even though the lock was released by the holding transaction. This fix is due to great work by Rich Prohaska. This problem is actually not specific to replication, normal transactions on a master will experience it too. But it hurts replication a lot, since replication must commit transactions in order, so one stalled transaction stalls all following transactions as well. 2. Implement the conflict detection and handling necessary for optimistic parallel replication to work. This basically implements thd_rpl_deadlock_check() and hton->kill_query methods. This should solve the problems where optimistic parallel replication with TokuDB breaks with lock wait timeouts. If someone wants to test it, I have made a tree available with just these fixes on top of MariaDB 10.1.19: https://github.com/knielsen/server/tree/tokudb_optimistic_parallel_replicat… The fix should appear in 10.1.20 eventually. The first part of the patch has also been submitted by Rich to upstream. When this is (hopefully) merged upstream, and upstream merged into MariaDB, the MariaDB version of the fix should be replaced with the Percona one. I tried making the MariaDB version of the fix identical to Rich's pull request and keeping it in a separate commit, so this should hopefully be simple to do when the time comes. - Kristian.

3 9

Re: [Maria-developers] [Commits] 00f5065: MDEV-11422 rpl.rpl_parallel_optimistic_nobinlog failed
by Jan Lindström 08 Dec '16

08 Dec '16

Hi, All other looks correct, I'm concerned only following change that is not needed to fix this bug: On Thu, Dec 8, 2016 at 2:35 PM, <marko.makela(a)mariadb.com> wrote: > > > - err = lock_trx_handle_wait(trx, > - (lock_mutex_taken || already_have_lock_mutex), > - (trx_mutex_taken || already_have_trx_mutex)); > +#ifdef UNIV_DEBUG > + dberr_t err = > +#endif > + lock_trx_handle_wait(trx, true, true); > + > + ut_ad(err == DB_SUCCESS || err == DB_LOCK_WAIT > + || err == DB_DEADLOCK); > > > In above is true, true always correct? we conditionally take lock and trx mutexes on some cases if we already do not have them, and below function will take them if they are not already taken. There is complex rules as Galera can call this from different situations and we already have had cases where some mutex was taken twice or no mutex was taken. R: Jan

1 0

[Maria-developers] MDEV-10787 Assertion `ltime->neg == 0' failed in void date_to_datetime(MYSQL_TIME*)
by Alexander Barkov 07 Dec '16

07 Dec '16

Hello Sergei, please review a patch for MDEV-10787. Thanks!

2 2

[Maria-developers] Temporary push of multi_range_read support to 10.2
by Michael Widenius 07 Dec '16

07 Dec '16

Hi! Kentoku, I have now pushed into 10.2 the code for spider patches: 014,015,023,033,035,037,040,042,044,045,049,050,051,053,059 The push is done to make it easier for Jacob and you to make test cases for the new code. I will work on reviewing and cleaning up this patch during the next few days and then do a push a final version when I have got some more test cases that covers most of the new code. Regards, Monty

1 0