- developers - lists.mariadb.org

Re: [Maria-developers] Rev 4006: MDEV-5674 - Performance: my_hash_sort_bin is called too often
by Sergei Golubchik 04 Mar '14

04 Mar '14

Hi, Sergey! On Feb 14, Sergey Vojtovich wrote: > At lp:maria/10.0 > > ------------------------------------------------------------ > revno: 4006 > revision-id: svoj(a)mariadb.org-20140214122541-tenvcllk760deai8 > parent: svoj(a)mariadb.org-20140213071355-psk0jpsarx2u7myc > committer: Sergey Vojtovich <svoj(a)mariadb.org> > branch nick: 10.0 > timestamp: Fri 2014-02-14 16:25:41 +0400 > message: > MDEV-5674 - Performance: my_hash_sort_bin is called too often > > - reduced number of my_hash_sort_bin() calls from 4 to 2 per query. > Let MDL subsystem use pre-calculated hash value for hash > inserts and deletes. Is this necessary? One of the reasons why I didn't review this your first patch, because I believed that the second one obsoletes it. Why do you need my_hash_insert_using_hash_value() when you achieve the same effect with your calc_hash callback in the second patch? > - reduced number of memory accesses done by my_hash_sort_bin() I'd expect the compiler to handle it automatically, but feel free to change it, if it doesn't. Regards, Sergei

2 1

Re: [Maria-developers] MDEV-5674 - Performance: my_hash_sort_bin is called too often
by Sergei Golubchik 04 Mar '14

04 Mar '14

Hi, Sergey! On Feb 20, Sergey Vojtovich wrote: > At lp:maria/10.0 > > ------------------------------------------------------------ > revno: 4007 > revision-id: svoj(a)mariadb.org-20140220072533-ko14y1714ouvd2f7 > parent: svoj(a)mariadb.org-20140214122541-tenvcllk760deai8 > committer: Sergey Vojtovich <svoj(a)mariadb.org> > branch nick: 10.0 > timestamp: Thu 2014-02-20 11:25:33 +0400 > message: > MDEV-5674 - Performance: my_hash_sort_bin is called too often It's MDEV-5675, not MDEV-5674 > Reuse MDL hash value in table cache: > - MDL namespace is excluded from hash value calculation, so that > hash value can be used by table cache as is > - hash value for MDL is calculated as resulting hash value + MDL > namespace > - extended hash implementation to accept user defined hash function This looks ok. I have a couple of comments, but they are mostly a question of taste. Up to you, you can push as is too. > === modified file 'sql/mdl.cc' > --- a/sql/mdl.cc 2014-02-14 12:25:41 +0000 > +++ b/sql/mdl.cc 2014-02-20 07:25:33 +0000 > @@ -760,13 +760,22 @@ void MDL_map::init() > } > > > +my_hash_value_type mdl_hash_function(const CHARSET_INFO *cs, > + const uchar *key, size_t length) > +{ > + return *((my_hash_value_type*) (key - offsetof(MDL_key, m_ptr) + > + offsetof(MDL_key, m_hash_value))) + > + key[0]; I'd prefer to avoid pointer and offset juggling. Because you use a custom hash function anyway, you can directly pass MDL_key as your "key", then here you can use ((MDL_key*)key)->hash_value() > +} > + > /** Initialize the partition in the container with all MDL locks. */ > > === modified file 'sql/mdl.h' > --- a/sql/mdl.h 2014-02-14 12:25:41 +0000 > +++ b/sql/mdl.h 2014-02-20 07:25:33 +0000 > @@ -397,7 +396,11 @@ class MDL_key > { > return & m_namespace_to_wait_state_name[(int)mdl_namespace()]; > } > - ulong hash_value() const > + my_hash_value_type hash_value() const > + { > + return m_hash_value + m_ptr[0]; I'm not sure it's a good idea, + is not a very good hash function :) On the other hand, I don't see anything immediately wrong with it in this particular case. Ok, let it be. Anyway, could you please use mdl_namespace() instead of m_ptr[0] where possible? > @@ -405,12 +408,14 @@ class MDL_key > private: > MDL_key(const MDL_key &); /* not implemented */ > MDL_key &operator=(const MDL_key &); /* not implemented */ > + friend my_hash_value_type mdl_hash_function(const CHARSET_INFO *, > + const uchar *, size_t); You probably wouldn't need this friend, if you'd change mdl_hash_function() to use MDL_key::hash_value() Regards, Sergei

2 5

[Maria-developers] Suggestion ,a new member
by anupama vijjapu 04 Mar '14

04 Mar '14

Hello fellow developers, I joined MariaDB community today and need few suggestions/projects/bugs that i can work on. I do not have any knowledge of MariaDB source, but I will start looking into it now. I have hands on experience in C/C++/Java/Python and MySQL Database. Any suggestions/recommendation are highly appreciated.

2 1

Re: [Maria-developers] Rev 4029: MDEV-4856 SQL_ERROR_LOG shows 1146 errors which didnt appear in mysql client
by Sergei Golubchik 03 Mar '14

03 Mar '14

Hi, Holyfoot! On Jan 31, holyfoot(a)askmonty.org wrote: > revno: 4029 > revision-id: holyfoot(a)askmonty.org-20140131103303-1rx7ielo83f8iahe > parent: holyfoot(a)askmonty.org-20140124020722-dd5twtwlcc8o1xiy > committer: Alexey Botchkov <holyfoot(a)askmonty.org> > branch nick: mdev-4856 > timestamp: Fri 2014-01-31 14:33:03 +0400 > message: > MDEV-4856 SQL_ERROR_LOG shows 1146 errors which didnt appear in mysql client. > The fill_schema_table() function used to call get_table_share() for > a table name in WHERE then clear the error list. That way plugins > receive the superfluous error notification if it happens in it. Also > the problem was that error handler didn't prevent the suppressed > error message from logging anyway as the logging happens in > THD::raise_condition before the handler call. > Fixed by adding the No_table_error_handler before the call. > raise_condition() also fixed. > === modified file 'sql/sql_class.cc' > --- a/sql/sql_class.cc 2014-01-23 18:21:02 +0000 > +++ b/sql/sql_class.cc 2014-01-31 10:33:03 +0000 > @@ -1145,7 +1145,6 @@ MYSQL_ERROR* THD::raise_condition(uint s > got_warning= 1; > break; > case MYSQL_ERROR::WARN_LEVEL_ERROR: > - mysql_audit_general(this, MYSQL_AUDIT_GENERAL_ERROR, sql_errno, msg); > break; > default: > DBUG_ASSERT(FALSE); > @@ -1156,6 +1155,8 @@ MYSQL_ERROR* THD::raise_condition(uint s > > if (level == MYSQL_ERROR::WARN_LEVEL_ERROR) > { > + mysql_audit_general(this, MYSQL_AUDIT_GENERAL_ERROR, sql_errno, msg); > + This is good. > === modified file 'sql/sql_show.cc' > --- a/sql/sql_show.cc 2013-11-19 12:16:25 +0000 > +++ b/sql/sql_show.cc 2014-01-31 10:33:03 +0000 > @@ -4199,8 +4199,13 @@ static int fill_schema_table_from_frm(TH > key_length= create_table_def_key(thd, key, &table_list, 0); > hash_value= my_calc_hash(&table_def_cache, (uchar*) key, key_length); > mysql_mutex_lock(&LOCK_open); > - share= get_table_share(thd, &table_list, key, > - key_length, OPEN_VIEW, &not_used, hash_value); > + { > + No_such_table_error_handler no_such_table_handler; > + thd->push_internal_handler(&no_such_table_handler); > + share= get_table_share(thd, &table_list, key, > + key_length, OPEN_VIEW, &not_used, hash_value); > + thd->pop_internal_handler(); > + } But I don't understand that. Do you mean, old code did *not* suppress errors here? How could selects from I_S tables work without it? Regards, Sergei

2 5

[Maria-developers] New member, Need suggestion
by Neil Shah 03 Mar '14

03 Mar '14

Hello fellow developers, I joined MariaDB community today and need few suggestions/projects/bugs that i can work on. I do not have any knowledge of MariaDB source, but I will start looking into it now. I have hands on experience in C/C++/Java/Python and MySQL Database. Any suggestions/recommendation are highly appreciated. -- Thanks & Regards Neil Shah

2 3

[Maria-developers] slave_ddl_exec_mode and incompatible change in MariaDB 10.0.8
by Pavel Ivanov 03 Mar '14

03 Mar '14

Hi, I've discovered a change in MariaDB 10.0.8 that I don't quite understand motivation for and which looks really surprising to me. So I would appreciate if you could tell me what's the motivation and why you think it's appropriate to do that. The change in question is http://bazaar.launchpad.net/~maria-captains/maria/10.0/revision/3961.1.7. I see several issues with it: 1. It introduces a pretty significant variable slave_ddl_exec_mode which wasn't announced in the Release Notes. 2. The default variable value was set to IDEMPOTENT which changes MariaDB slave behavior in an incompatible way. Not everyone will want to have this variable set to IDEMPOTENT (I'd actually argue that no one should set it to IDEMPOTENT), but without any mentioning of this variable in the Release Notes people basically have the only chance to get knowledge about this change (let me stress out: _incompatible_ change) from production incidents or from accidental code inspection. 3. When slave_ddl_exec_mode is set to IDEMPOTENT every "DROP TABLE" event in master's binlog is converted into "DROP TABLE IF EXISTS" in slave's binlog, which I believe is a major no-no for replication. Binlogs on master and slave should be identical, otherwise after failover even if one would set slave_ddl_exec_mode = STRICT it will still execute "DROP TABLE IF EXISTS" because new master has that in binlogs. 4. Why this change introduces OPTION_GTID_BEGIN in the code? Why OPTION_BEGIN wasn't enough? From reading the code the only thing I understand is that with broken binlogs and sequence of events "GTID BEGIN; CREATE TABLE; CREATE TABLE" MariaDB will try to execute both CREATE TABLE statements in the same transaction without auto-committing at the end. This sounds pretty crazy to me. And given the fact that in normal situation such sequence of events on the master is not possible it seems that existence of both OPTION_BEGIN and OPTION_GTID_BEGIN only introduces confusion (MariaDB have GTID always on, right? So it should always execute GTID_BEGIN and never BEGIN without GTID, right?). So what is your word on this? Can you at least set default value for slave_ddl_exec_mode to STRICT so that slave behavior didn't change with the version upgrade? Thank you, Pavel

2 7

Re: [Maria-developers] MariaDB Master Replication To MySQL
by Kristian Nielsen 03 Mar '14

03 Mar '14

> In this particular case, we have someone that wants to use MariaDB 10.0 master with a MySQL 5.6 slave. But they are running into a problem with GTID. Generally, the recommended setup is to only replicate from old version on master to new (or same) version on slave. Replicating from new master to old slave in general is hard to support 100%, as there may be eg. new SQL syntax on the master that is impossible for the old version slave to execute. That being said, the intention is that replication from new master to old slave (like MariaDB 10.0 -> MySQL 5.6) should work in most reasonable cases. So this sounds like a bug which might be worth filing. > 2.) Is it possible to disable GTID in 10.0? It seems like it would break a few other things like parallel replication, but they're only running MariaDB on the master. GTID is disabled automatically when the slave does not support it. The code rewrites the binlog stream on the fly to remove events that are not supported on the old slave, and replace them with something appropriate for that slave version. Does the same problem occur when replicating from MariaDB 10.0 master to MariaDB 5.5 slave? What about when replicating from MySQL 5.5 master to MySQL 5.6 slave? - Kristian.

1 0

Re: [Maria-developers] MDEV-5755 - Scalability: let LOCK_open and THR_LOCK::mutex be normal mutexes
by Sergei Golubchik 03 Mar '14

03 Mar '14

Hi, Sergey! On Feb 27, Sergey Vojtovich wrote: > At lp:maria/10.0 > > ------------------------------------------------------------ > revno: 4011 > revision-id: svoj(a)mariadb.org-20140227141521-fgj5b0h3qbijyojh > parent: svoj(a)mariadb.org-20140220072533-l3au3y7mrsyclfpb > committer: Sergey Vojtovich <svoj(a)mariadb.org> > branch nick: 10.0 > timestamp: Thu 2014-02-27 18:15:21 +0400 > message: > MDEV-5755 - Scalability: let LOCK_open and THR_LOCK::mutex be normal mutexes > > Normal mutexes scale better than adaptive at high thread count. > At low thread count they behave similarly. To summarize Axel results: at low concurency normal can be ~1% slower. at high concurrency r/o - normal are up to 25% faster. at high concurrency r/w - normal are up to 10% slower. That last result bothers me a litte. Regards, Sergei

2 1

[Maria-developers] MDEV-5081 - Simple performance improvement for MariaDB
by Sergey Vojtovich 28 Feb '14

28 Feb '14

Hi Sergei, do you know if we ship any binaries with fast mutex (-DMY_PTHREAD_FASTMUTEX) enabled or officially suggest it? What was it's intention and what loads it was supposed to make faster? I'm trying to understand if fastmutex is worth fixing. It looks like duplication of PTHREAD_MUTEX_ADAPTIVE_NP, so we have double spinning when they're enabled. According to my recent benchmarks it shown worst performance. Thanks, Sergey

3 5

[Maria-developers] Keystone authentication plugin GSoc 14
by Abhishek Solanki 28 Feb '14

28 Feb '14

Hello MariaDB members , I am interested to work on "Keystone authentication plugin" project , i have a good knowledge of C/C++ and OpenStack. So can anyone help me how to get started with this project, so that i cold dig something before applying for this project. -- Thanks, Abhishek

2 1