September 2017 - developers - lists.mariadb.org

[Maria-developers] Windows line endings in MariaDB sources again (
by Otto Kekäläinen 24 Jul '19

24 Jul '19

Hello! While trying to release 10.0.26 for Debian and Ubuntu I noticed git chokes again on Windows line endings. As our .gitattributes states that by default git should clean up the line endings, why do we keep getting new Windows line endings in new code? mariadb-10.0$ cat .gitattributes # Set the default behavior, in case people don't have core.autocrlf set. * text=auto ... storage/connect/mysql-test/connect/std_data/*.txt -text mariadb-10.0.26$ find . | xargs file | grep CRLF ./storage/connect/mysql-test/connect/std_data/boyswin.txt: ASCII text, with CRLF line terminators ./storage/connect/mysql-test/connect/std_data/expenses.txt: ASCII text, with CRLF line terminators ./storage/connect/mysql-test/connect/std_data/emp.txt: ASCII text, with CRLF line terminators ./mysql-test/std_data/loaddata7.dat: ASCII text, with CRLF line terminators ./mysql-test/r/loadxml.result: ASCII text, with CRLF, LF line terminators ./mysql-test/r/perror-win.result: ASCII text, with CRLF, LF line terminators ./mysql-test/r/mysql_binary_mode.result: ASCII English text, with CRLF, LF line terminators ./mysql-test/r/func_regexp_pcre.result: UTF-8 Unicode C++ program text, with CRLF, CR, LF line terminators ./pcre/testdata/grepoutputN: ASCII text, with CRLF, CR, LF line terminators ./pcre/testdata/greppatN4: ASCII text, with CRLF line terminators This leads for me to: dpkg-source: info: building mariadb-10.0 using existing ./mariadb-10.0_10.0.26.orig.tar.gz dpkg-source: info: local changes detected, the modified files are: mariadb-10.0/storage/connect/mysql-test/connect/std_data/boyswin.txt mariadb-10.0/storage/connect/mysql-test/connect/std_data/emp.txt mariadb-10.0/storage/connect/mysql-test/connect/std_data/expenses.txt dpkg-source: error: aborting due to unexpected upstream changes, see /tmp/mariadb-10.0_10.0.26-1.diff.Yzwwhw Which means I have to do extra work with importing the upstream sources.. - Otto

2 4

[Maria-developers] 10.2 and merges (e.g. TokuDB)
by Phil Sweeney 28 Aug '18

28 Aug '18

Hi there, It seems 10.2.6 is still shipping with TokuDB from PS 5.6.35-80 (Feb 8th 2017). I can't see a jira issue for a 10.2 merge since 10.2.4. Perhaps intentional to help stabilise 10.2 for release? 10.0.30 also shipped with the same Feb version (in March). 10.1.23 however was released in May and still has this version from Feb, which seems unfortunate, but I know it's because it has to flow through 10.0 first. I see 10.0.31 is due to get TokuDB from PS 5.6.36-82 And that should flow into 10.1.24 soon too. However, 10.2.7 isn't due until July, so by then TokuDB version will be 5 months old in 10.2 I'm waiting on a fairly important row count fix that should appear in the next Percona Server release. When it does, if the current merging strategy continues, do I have to first wait for 10.0.32 to get it in late July, then 10.1.26 in August, then 10.2 after that? Or with 10.0, 10.1 and 10.2 all being supported and needing merges, is there a new strategy on the cards to shorten this lag? Thanks! Phil

3 7

[Maria-developers] MDEV-13118 Wrong results with LOWER and UPPER and subquery
by Alexander Barkov 12 Jul '18

12 Jul '18

Hello Sergei, Please review a patch for MDEV-13118. Thanks!

2 4

Re: [Maria-developers] Mdev-10664 Add statuses about optimistic parallel replication stalls.
by Sachin Setiya 19 Apr '18

19 Apr '18

Hi Kristian! Sorry for late reply , I was on vacations. On Tue, Jan 10, 2017 at 7:31 PM, Kristian Nielsen <knielsen(a)knielsen-hq.org> wrote: > Sachin Setiya <sachin.setiya(a)mariadb.com> writes: > > >> Right, but you still need _some_ way to synchronise multiple threads > > > I used atomic builtins my_atomic_add64_explicit. > > That seems fine (but see comment below). > > > Documentation:- > > > > Slave_DDL_Events:- > > > > Description: Number of DDL statements slave have executed in > > replication channel X. > > Scope: Per Replication Channel > > Data Type: numeric > > Introduced: MariaDB 10.1 > > > > Slave_Non_Transactional_Events:- > > > > Description: Number of Non Transactional "no idea what should be the > > next word " slave have executed in replication channel X. > > Scope: Per Replication Channel > > Data Type: numeric > > Introduced: MariaDB 10.1 > > I checked, the term "event group" is used already multiple places in the > documentation, so I think it can be used here also. An event group can be a > transaction (possibly with multiple transactional DML statements), or a DDL > or non-transactional statement. > > What you are doing is introducing status variables to help understand > optimistic (and aggressive) parallel replication modes. > > In optimistic parallel replication, a non-transactional event group is not > safe to run in parallel with other non-transactional events. So that event > is made to wait for all prior event groups to reach commit state, before > being allowed to execute. However, other transactional events are not > delayed by the non-transactional event group. > > Slave_Non_Transactional_Events counts the occurences of this. > > Then in case of a DDL statement, this can potentially be unsafe to run in > parallel even with other transactional events. So for a DDL, the DDL has to > wait for all prior event groups, _and_ subsequent event groups have to wait > for the DDL. > > Slave_DDL_Events counts these occurences. > > This is what the documentation, and commit messages, should explain, I > think. > > I think the status variables describe the values for > @@default_master_connection?, this should also be explained in the > documentation. > > Why did you decide to put this information into a status variable? > Normally, > slave status is seen in SHOW SLAVE STATUS, which supports showing status > for > a particular connection without using @@default_master_connection. > > Sorry for this , I guess It was a confusion I was looking at Slave_Running this is avaliable in show status as well as show slave status. So I thought for if I want to show some information in SHOW slave status this has to be in show status. And this is wrong. Now we can see non trans events / ddl events in show slave status. Show I remove this vars from Show Status, or it is okay ? > > diff --git a/sql/log_event.h b/sql/log_event.h > > index 90900f6..579607f 100644 > > --- a/sql/log_event.h > > +++ b/sql/log_event.h > > @@ -50,6 +50,7 @@ > > #endif > > > > #include "rpl_gtid.h" > > +#include "my_atomic.h" > > Why include my_atomic.h in the header file, when you only use it in the .cc > file? > Changed. > > > + mysql_mutex_lock(&LOCK_active_mi); > > + if (master_info_index) > > + { > > + mi= master_info_index-> > > + get_master_info(&thd->variables.default_master_connection, > > + Sql_condition::WARN_LEVEL_NOTE); > > + if (mi) > > + tmp= mi->total_ddl_events; > > > + if (mi) > > + tmp= mi->total_non_trans_events; > > Better use an atomic primitive here to read the value. Taking a mutex does > not help synchronise against my_atomic_add64_explicit(). > > Changed. > > + } > > + mysql_mutex_unlock(&LOCK_active_mi); > > + if (mi) > > + *((longlong *)buff)= tmp; > > + else > > + var->type= SHOW_UNDEF; > > + > > + return 0; > > +} > > #endif /* HAVE_REPLICATION */ > > > > static int show_open_tables(THD *thd, SHOW_VAR *var, char *buff, > > > index 9365c06..a474aa5 100644 > > --- a/sql/rpl_mi.h > > +++ b/sql/rpl_mi.h > > @@ -303,6 +303,12 @@ class Master_info : public > Slave_reporting_capability > > > > /* The parallel replication mode. */ > > enum_slave_parallel_mode parallel_mode; > > + > > + /* No of DDL statements */ > > + long total_ddl_events; > > + > > + /* No of non-transactional statements*/ > > + long total_non_trans_events; > > "long" is not the appropriate type here. Certainly not since you are using > my_atomic_add64_explicit(). Use a type that is the same size on all > platforms (and why use a signed type?). > > Changed to uint64. > - Kristian. > Attached a newer version of patch , please review it. -- Regards Sachin

3 7

[Maria-developers] CREATE SEQUENCE is coming
by Michael Widenius 18 Apr '18

18 Apr '18

Hi! For those that are trying to port applications to MariaDB from other databases or need more features than the current AUTO_INCREMENT provides, the following may be of interest: First version of sequence is pushed to bb-10.2-sequence. It should be merged to bb-10.2-compatilbity and then to 10.3 shortly (1-2 weeks). What is working for the moment: CREATE OR REPLACE [TEMPORARY] SEQUENCE [IF NOT EXISTS] name [ INCREMENT [ BY | = ] increment ] [ MINVALUE [=] minvalue | NO MINVALUE ] [ MAXVALUE [=] maxvalue | NO MAXVALUE ] [ START [ WITH | = ] start ] [ CACHE [=] cache ] [ [ NO ] CYCLE ] ENGINE=xxx COMMENT=".." SELECT NEXT VALUE FOR sequence_name; SELECT NEXTVAL(sequence_name); SELECT PREVIOUS VALUE FOR sequence_name; SELECT LASTVAL(sequence_name); SHOW CREATE SEQUENCE sequence_name; SHOW CREATE TABLE sequence_name; CREATE TABLE sequence-structure ... SEQUENCE=1 ALTER TABLE sequence RENAME TO sequence2; RENAME TABLE sequence RENAME TO sequence2; DROP [TEMPORARY] SEQUENCE sequence_name [IF EXISTS] See https://jira.mariadb.org/browse/MDEV-10139 for progress. See commit messages in bb-10.2-sequence for what is still to be done. Some documentation can be found at: https://mariadb.com/kb/en/mariadb/create-sequence/ This will be improved as the work progress. One of the goals with the SEQUENCE implementation is that all old tools, like mysqldump, should work unchanged, while still keeping normal usage of sequence standard compatibly. The make this possible, the sequence is currently implemented as a table with a few exclusive properties. The main disadvantage of having sequence as a table is that it uses the same name space as tables. In other words, you can't have a table and a sequence of the same name. The benefit is that sequences shows up in 'show tables', one can create a sequence also with 'create table' and drop it with 'drop table'. One can select from it as from any other table. This ensures that all old tools that works with tables should work with sequences. The special properties for sequence tables are: - A sequence table has always one row. - When one creates a sequence, either with CREATE TABLE or CREATE SEQUENCE, one row will be inserted. - Normal table options works for CREATE SEQUENCE. One can use ENGINE=xxx, COMMENT=xxx etc. - If one tries to insert into a sequence table, the single row will be updated. This allows mysqldump to work but also gives the additional benefit that one can change all properties of a sequence with a single insert. New applications can of course also use ALTER SEQUENCE. - Updates to the sequence table will change the single row. - Doing a select on the sequence shows the current state of the sequence, except the values that are reserved in the cache. The column 'next_value' shows the next value not reserved by the cache. - Truncate on a sequence table will give an error. - Alter table and rename works on sequences. - If one creates a sequence with INCREMENT 0, then the sequence will use auto_increment_increment and auto_increment_offset for the sequence, just like AUTO_INCREMENT. This allows sequences to work reliable in a master-master environment and with Galera. Internally sequence tables are created as a normal table without rollback (InnoDB, Aria and MySAM supports this) with is wrapped by a sequence engine. This allowed me to create sequences with almost no performance impact for normal tables. (The cost is one 'if' per insert if binary log is enabled). MariaDB 10.3 will support both the ANSI SQL and Oracle syntax for creating and accessing sequences. As ANSI SQL doesn't have an easy access to the last generated value, 103 also supports 'PREVIOUS VALUE FOR sequence_name', like IBM DB2, and LASTVAL(sequence_name) as PostgreSQL. I want to thank Jianwe Zhao from Aliyun for his work on SEQUENCE in AliSQL which gave me ideas and inspiration for this work. I also want to thank Marko Mäkelä for his help in the InnoDB part of the code. There is still a lot of work to fix edge cases, but in the current implementation most major things seams to work... Comments, suggestions or questions? Regards, Monty

2 1

[Maria-developers] a doubt about locks on supremum
by hongxiang jiang 18 Jan '18

18 Jan '18

2 1

[Maria-developers] Add hidden level 2/3 (PSEDUO_HIDDEN_COLUMN / COMPLETELY_HIDDEN_COLUMN) to mariadb
by Sachin Setiya 27 Nov '17

27 Nov '17

Hello Everyone! Currently I am implementing 10177. But I have one doubt I am not sure how to implement a function which will create hidden level 2/3 field for server. Suppose we create a function (say add_hidden_field(.....)) which will add the field to table T. But the issue is where should we call this function in mysql_prepare_create_table / mysql_prepare_alter_table or init_from_binary_frm_image Doing it from init_from_binary_frm_image make sense because this way server and slave frm of table T will be same (We have decided that we wont log hidden level 2/3 field to binlog) , But the issue is optimizer wont be ready to decide whether we require a hidden level 3 field at that time. We can also do it from mysql_prepare_create_table , but I dont think this wont be good idea because at this time optimizer wont be ready plus this will complicate binlog handling. Or may be we should have a function which can modify table object any time. But I am not sure if it is possible/good idea ? What you guys think ? -- Regards Sachin Setiya Software Engineer at MariaDB

2 4

[Maria-developers] Please review MDEV-11292 Item_func_sp::fix_length_and_dec() does not set collation derivation properly
by Alexander Barkov 20 Nov '17

20 Nov '17

Hello Alexey, Please review a patch for MDEV-11292 (for 10.3). Thanks.

2 2

[Maria-developers] Obsolete GTID domain delete on master (MDEV-12012, MDEV-11969)
by andrei.elkin＠pp.inet.fi 24 Oct '17

24 Oct '17

Hello. Let me propose methods to clean master off unused gtid domains. The issue is quite practical, as a couple of references on the subject line tell. Either of them narrates a scenario of two default domain-id masters serving to one slave initially by "legacy" non-gtid protocol. When later the masters have changed their common domain to different private ones *and* the slave turns gtid replication ON, it can't connect to a master due to the gtid protocol. And the reason is the prior-to-gtid-connect gained slave's gtid position consisting of gtids generated by the other master obviously (the two masters never replicated each other) does not fit to the current master binlog state. The past default domain-id is actually permanent past from the user perspective in these cases. Its events have been already replicated and none new will be generated and replicated. Therefore such domain conceptually may be cleaned away from either the masters and slave states. Once it's done, the gtid-enabled slave will successfully connect to any master. The slave state purge is simple SET @@global.gtid_slave_pos to a sequence free of the purged domain. The master side binlog state one requires a new "command" unless the user is happy with RESET MASTER. While setting the new gtid binlog state to be old-domain-free we would like for the new "command" to preserve the existing binlog files. This could be accomplished as Kristian suggests in MDEV-12012 (I could not find any earlier references to the idea) as a new option to FLUSH BINARY LOGS DELETE DOMAIN d1, d2 KN> This command would check that the current binlog state is equal to KN> the GTID_LIST_LOG_EVENT at the start of the first binary log file, KN> within the specified domains. If not, it is an error. But if so, KN> the new binary log file created by FLUSH will be written with the KN> specified domains omitted from GTID_LIST_LOG_EVENT (and the current KN> binlog state updated accordingly). The idea looks quite sane, I only could not grasp why presence of being deleted domains in the very first binlog's GTID_LIST_LOG_EVENT list is warrant for throwing error. Maybe we should leave it out to the user, Kristian? That is to decide what domain is garbage regardless of the binlog state history. While the FLUSH way looks sufficient and robust I could not help to think over an alternative. Consider a scenario when a domain's sequence number got run out of range. While deems unrealistic in practice we can simulate it with SET @@SESSION.gtid_domain_id=11; SET @@SESSION.server_id=1; SET @@SESSION.gtid_seq_no=18446744073709551615; /* Exec some dummy loggable query, e.g */ CREATE TABLE IF NOT EXISTS `table_dummy`; SHOW LOCAL VARIABLES LIKE '%gtid_binlog_pos%'; 11-1-18446744073709551615 SET @@SESSION.gtid_seq_no=0 DROP TABLE `table_dummy`; SHOW LOCAL VARIABLES LIKE '%gtid_binlog_pos%'; 11-1-0 I've used two gtids to show domain overflow because I also liked to read the zero of the last gtid as ... there's *nothing* in this domain. So it's actually a new "namesake" one, replacing the old that is wrapped around. The 1st group of events created in the new domain - 11-1-1 - could shadow the old domain's 11-1-1 as well as all the rest of the old domain from gtid replication protocol. And that means the old domain is actually deleted. So if my reading of zero is correct the binlog status would be empty instead. That's how we also can approach the master side old gtid domain purging: 1. Leave wrapping around an old domain to the user via running the queries like above; 2. The binary logger would be made to react on the fact of wrap-around with binary log rotation ("internal" FLUSH BINARY LOG). And the new binlog file won't contain the wrapped "away" domain (because there are no new event group in it of yet). I would be glad to hear your opinions, dear colleagues. Cheers, Andrei

4 27

[Maria-developers] MDEV-13417 / MDEV - 13418 The order of evaluation of SELECT..INTO assignments
by jerome brauge 03 Oct '17

03 Oct '17

Hello Sergei, Here is a patch for MDEV-13417 and MDEV-13418. Until you choose the right name for this new sql_mode, I temporarily chosen "SET_CONSISTENCY". It is set only for sql_mode=oracle for now. Can you review it ? Thank you. Regard.

2 6