- developers - lists.mariadb.org

[Maria-developers] New (by Knielsen): Efficient group commit for binary log (116)
by worklog-noreply＠askmonty.org 26 Apr '10

26 Apr '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Efficient group commit for binary log CREATION DATE..: Mon, 26 Apr 2010, 13:28 SUPERVISOR.....: Knielsen IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-RawIdeaBin TASK ID........: 116 (http://askmonty.org/worklog/?tid=116) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: Currently, in order to ensure that the server can recover after a crash to a state in which storage engines and binary log are consistent with each other, it is necessary to use XA with durable commits for both storage engines (innodb_flush_log_at_trx_commit=1) and binary log (sync_binlog=1). This is _very_ expensive, since the server needs to do three fsync() operations for every commit, as there is no working group commit when the binary log is enabled. The idea is to - Implement/fix group commit to work properly with the binary log enabled. - (Optionally) avoid the need to fsync() in the engine, and instead rely on replaying any lost transactions from the binary log against the engine during crash recovery. For background see these articles: http://kristiannielsen.livejournal.com/12254.html http://kristiannielsen.livejournal.com/12408.html http://kristiannielsen.livejournal.com/12553.html ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Windows test platforms
by Bo Thorsen 26 Apr '10

26 Apr '10

Hi everyone, I'm going to build a set of Windows installations to run buildbot on. But finding the right set of versions to do this is a challenge. There are so many versions of Windows that it's impossible to get even close to the full set of them. So I'd like to have a good set of systems that pretty much cover most options. The buildbot system will work in three parts. One will build a set of installable packages - for now probably just a zip file. Another will install these packages and test them on clean systems. A third will do compilation tests on OS/Compiler combinations. Windows 2000 or earlier is not going to be done. I'm not sure about Windows Server 2003. I'd like to skip XP, as MS cut support for it, but I think there are too many live installations to do this yet. For building the installable packages, I intend to use XP and Visual Studio 2008. My list of systems to test the packages and do the compile + make test on would be: XP Pro 32 bit Vista 32 bit Windows 7 64 bit Server 2008 64 bit On XP, I'd install VC++ 2005, on the others VC++ 2008. When Visual Studio 2010 is out, I'll install that on Vista and one of the 64 bit systems. This is four systems that does a very thin span of the incredible amount of systems done by MS. (And people say there are a lot of Linux distributions.) Does this make sense to you? Please let me know what you think. Kind regards, Bo Thorsen, Monty Program AB.

2 1

Re: [Maria-developers] Windows test platforms
by Bo Thorsen 26 Apr '10

26 Apr '10

Monday 26 April 2010 11:04:50 Sergey Petrunya wrote: > On Mon, Apr 26, 2010 at 07:52:55AM +0200, Bo Thorsen wrote: > > The buildbot system will work in three parts. One will build a set of > > installable packages - for now probably just a zip file. Another will > > install these packages and test them on clean systems. A third will do > > compilation tests on OS/Compiler combinations. > > And none of them will run mysql-test-run? Why? I didn't write it, but I assume both the test platforms will run the test cases. The package build won't. Is this what you expected? Kind regards, Bo Thorsen, Monty Program AB.

1 0

Re: [Maria-developers] Windows test platforms
by Bo Thorsen 26 Apr '10

26 Apr '10

Monday 26 April 2010 11:25:53 Michael Widenius wrote: > Hi! > > >>>>> "Bo" == Bo Thorsen <bo(a)askmonty.org> writes: > > <cut> > > Bo> For building the installable packages, I intend to use XP and Visual > Studio Bo> 2008. > > We should look at using Visual Studio 2010 ASAP, as this is said to > give significantly better performance. The only reason I didn't choose VS2010 is that it's only beta at this moment. When it's out of beta, I agree we should use this instead. However, having a test buildbot system with the 2010 beta is probably a good idea, to get rid of compiler problems from it. > Bo> My list of systems to test the packages and do the compile + make test > on Bo> would be: > > Bo> XP Pro 32 bit > Bo> Vista 32 bit > Bo> Windows 7 64 bit > Bo> Server 2008 64 bit > > Bo> On XP, I'd install VC++ 2005, on the others VC++ 2008. When Visual > Studio 2010 Bo> is out, I'll install that on Vista and one of the 64 bit > systems. > > Bo> This is four systems that does a very thin span of the incredible > amount of Bo> systems done by MS. (And people say there are a lot of Linux > distributions.) > > Bo> Does this make sense to you? Please let me know what you think. > > Sounds good. > > We should probably start with trying to get packages for Windows 7, 64 > bit. I hope we can have just two packages - 32 and 64 bit - that will install on all Windows systems. There might be differences between the systems that have to be handled in the packages, but I don't want to have two packages per Windows version. That would be *really* bad. Kind regards, Bo Thorsen, Monty Program AB.

1 0

[Maria-developers] Storage Engine Summit 2010
by Sergei Golubchik 25 Apr '10

25 Apr '10

Hi, all! We are happy to announce the *Storage Engine Summit* 2010. This traditional one-day meeting of Storage Engine vendors/authors takes place yearly right after the User Conference. This year it will be held on Friday, April 16h. Invited are all MySQL/MariaDB Storage Engine vendors and anyone interested in the MySQL/MariaDB Storage Engine API. Please register in advance, the space is limited. We will talk about new developments in the area of the MySQL/MariaDB Storage Engine API - what was implemented recently, what we are working on - and about new directions of the development, what needs to be implemented to make the life of Storage Engine authors easier. See http://askmonty.org/wiki/Storage_Engine_Summit_2010 for details. Regards, Serg

2 2

[Maria-developers] New (by Serg): innodb statistics in the slow log (115)
by worklog-noreply＠askmonty.org 25 Apr '10

25 Apr '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: innodb statistics in the slow log CREATION DATE..: Sun, 25 Apr 2010, 16:35 SUPERVISOR.....: Sergei IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 115 (http://askmonty.org/worklog/?tid=115) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: We need to find some way for innoDB/XtraDB specific information to appear in the slow log. Same effect as in Percona patches, but a with cleaner implementation. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Serg): innodb statistics in the slow log (115)
by worklog-noreply＠askmonty.org 25 Apr '10

25 Apr '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: innodb statistics in the slow log CREATION DATE..: Sun, 25 Apr 2010, 16:35 SUPERVISOR.....: Sergei IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 115 (http://askmonty.org/worklog/?tid=115) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: We need to find some way for innoDB/XtraDB specific information to appear in the slow log. Same effect as in Percona patches, but a with cleaner implementation. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Serg): insert ignore ha_extra hint (114)
by worklog-noreply＠askmonty.org 25 Apr '10

25 Apr '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: insert ignore ha_extra hint CREATION DATE..: Sun, 25 Apr 2010, 16:32 SUPERVISOR.....: Sergei IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 114 (http://askmonty.org/worklog/?tid=114) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: There is HA_EXTRA_WRITE_CAN_REPLACE hint that tells the engine that the following ::write_row() calls are part of the REPLACE statement, not INSERT. With this knowledge the engine can execute the replace internally, deleting the conflicting row in the ::write_row() method instead of returning an error. We need a similar HA_EXTRA_WRITE_CAN_IGNORE hint to allow engines to optimize INSERT IGNORE in a similar way. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Serg): insert ignore ha_extra hint (114)
by worklog-noreply＠askmonty.org 25 Apr '10

25 Apr '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: insert ignore ha_extra hint CREATION DATE..: Sun, 25 Apr 2010, 16:32 SUPERVISOR.....: Sergei IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 114 (http://askmonty.org/worklog/?tid=114) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: There is HA_EXTRA_WRITE_CAN_REPLACE hint that tells the engine that the following ::write_row() calls are part of the REPLACE statement, not INSERT. With this knowledge the engine can execute the replace internally, deleting the conflicting row in the ::write_row() method instead of returning an error. We need a similar HA_EXTRA_WRITE_CAN_IGNORE hint to allow engines to optimize INSERT IGNORE in a similar way. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Serg): support many clustered keys per table (113)
by worklog-noreply＠askmonty.org 25 Apr '10

25 Apr '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: support many clustered keys per table CREATION DATE..: Sun, 25 Apr 2010, 16:23 SUPERVISOR.....: Sergei IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 113 (http://askmonty.org/worklog/?tid=113) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: The server now assumes that there can be only one clustered key per table, and only primary key can be clustered. It's not true for certain storage engines, we need to remove this limitation ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0