Comments on: More Gotchas with MySQL 5.0 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/ Everything about MySQL Performance Tue, 02 Dec 2008 15:43:09 +0000 http://wordpress.org/?v=2.5.1 By: pero on anything » Blog Archive » SHOW STATUS considered harmful http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-354431 pero on anything » Blog Archive » SHOW STATUS considered harmful Thu, 11 Sep 2008 15:16:21 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-354431 [...] Peter Zaitev points out calling SHOW STATUS might have a huge performance [...] [...] Peter Zaitev points out calling SHOW STATUS might have a huge performance [...]

]]> By: domas mituzas: vaporware, inc. » Blog Archive » Performance of status commands http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-305368 domas mituzas: vaporware, inc. » Blog Archive » Performance of status commands Mon, 26 May 2008 19:45:53 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-305368 [...] wrote a year ago about performance regression in ‘SHOW STATUS’. Of course, now that Mark has filed a bug [...] [...] wrote a year ago about performance regression in ‘SHOW STATUS’. Of course, now that Mark has filed a bug [...]

]]> By: Vasil Dimov http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-301282 Vasil Dimov Mon, 19 May 2008 15:42:21 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-301282 Patch that removes the variable Innodb_buffer_pool_pages_latched from SHOW STATUS has been committed in InnoDB 5.0 and 5.1, follow the bug report: http://bugs.mysql.com/36600 Patch that removes the variable Innodb_buffer_pool_pages_latched from SHOW STATUS has been committed in InnoDB 5.0 and 5.1, follow the bug report: http://bugs.mysql.com/36600

]]> By: peter http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299587 peter Fri, 16 May 2008 21:42:08 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299587 Oh yes indeed. Oh yes indeed.

]]> By: Vasil Dimov http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299534 Vasil Dimov Fri, 16 May 2008 19:39:50 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299534 Hmm, Peter, do you mean SHOW STATUS instead of SHOW INNODB STATUS? buf_get_latched_pages_number() is called when SHOW STATUS is executed, it is not called when SHOW INNODB STATUS is executed. I have prepared a simple patch that removes the Innodb_buffer_pool_pages_latched variable from SHOW STATUS. Hmm,

Peter, do you mean SHOW STATUS instead of SHOW INNODB STATUS?
buf_get_latched_pages_number() is called when SHOW STATUS is executed, it is not called when SHOW INNODB STATUS is executed.

I have prepared a simple patch that removes the Innodb_buffer_pool_pages_latched variable from SHOW STATUS.

]]> By: peter http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299528 peter Fri, 16 May 2008 18:10:43 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299528 Heikki, Thanks this is good change. I know you guys read bugs and listen to our feedback but it is not exactly the same as dealing with Innodb in production. Heikki,

Thanks this is good change.
I know you guys read bugs and listen to our feedback but it is not exactly the same as dealing with Innodb in production.

]]> By: peter http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299527 peter Fri, 16 May 2008 18:08:23 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299527 Vasil, I think it is not the time SHOW INNODB STATUS take which is important but the fact other queries starve waiting on innodb buffer pool which can be significant. I've seen hundreds of queries to be queued this way which then released at once and cause significant resource competition. Vasil,

I think it is not the time SHOW INNODB STATUS take which is important but the fact other queries starve waiting on innodb buffer pool which can be significant. I’ve seen hundreds of queries to be queued this way which then released at once and cause significant resource competition.

]]> By: Heikki Tuuri http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299515 Heikki Tuuri Fri, 16 May 2008 14:51:25 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299515 All, we are going to remove the loop to calculate buf_get_latched_pages_number from the standard build of mysqld. We may keep it in the UNIV_DEBUG build. Since SHOW INNODB STATUS\G prints all reserved latches, people can get the information, and in a much more detailed form, from SHOW INNODB STATUS\G. ### Concerning the "disconnectedness of InnoDB developers" from the MySQL user base, InnoDB developers look at bugs.mysql.com every day. And consulting companies like Percona exist, who give us feedback :). Regards, Heikki All,

we are going to remove the loop to calculate buf_get_latched_pages_number from the standard build of mysqld. We may keep it in the UNIV_DEBUG build.

Since SHOW INNODB STATUS\G prints all reserved latches, people can get the information, and in a much more detailed form, from SHOW INNODB STATUS\G.

###

Concerning the “disconnectedness of InnoDB developers” from the MySQL user base, InnoDB developers look at bugs.mysql.com every day. And consulting companies like Percona exist, who give us feedback :).

Regards,

Heikki

]]> By: Vasil Dimov http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299505 Vasil Dimov Fri, 16 May 2008 14:14:30 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-299505 Can someone give an estimate how much SHOW STATUS improves when buf_get_latched_pages_number() is disabled? Thanks! Can someone give an estimate how much SHOW STATUS improves when buf_get_latched_pages_number() is disabled?

Thanks!

]]> By: peter http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-296743 peter Thu, 08 May 2008 18:55:59 +0000 http://www.mysqlperformanceblog.com/2007/07/27/more-gotchas-with-mysql-50/#comment-296743 Mark, I'm not at all surprised :) Innodb developers are very disconnected from MySQL user base and customer base. I do not think there are guys which deal with large Innodb installations on daily basics in Innodb team and Innodb/Oracle does not have Support or Consulting teams which could work close with developers on issues like this. I mean of course people report bugs like this but I do not think their importance is felt by hart unless you deal with it in production. Mark,

I’m not at all surprised :)

Innodb developers are very disconnected from MySQL user base and customer base. I do not think there are guys which deal with large Innodb installations on daily basics in Innodb team and Innodb/Oracle does not have Support or Consulting teams which could work close with developers on issues like this.

I mean of course people report bugs like this but I do not think their importance is felt by hart unless you deal with it in production.

]]>