Comments for MySQL Performance Blog

Comment on The relationship between Innodb Log checkpointing and dirty Buffer pool pages by Jay Janssen

Jay Janssen — Wed, 22 Feb 2012 14:14:48 +0000

Shenglin, it’s pretty obvious when it starts happening, because your query throughput gets very bad.

Even on 5.1, you may consider using the Innodb plugin, which has some of these variables: http://dev.mysql.com/doc/innodb-plugin/1.0/en/innodb-parameters-new.html

http://dev.mysql.com/doc/innodb-plugin/1.0/en/innodb-plugin-installation-overview.html

Comment on Announcing MySQL Monitoring Plugins from Percona by snowave

snowave — Wed, 22 Feb 2012 07:24:31 +0000

so when do incremental backup ，the –slave-info couldn‘t be added to the command line？

Comment on Announcing MySQL Monitoring Plugins from Percona by snowave

snowave — Wed, 22 Feb 2012 07:13:16 +0000

hello,Baron Schwartz, i used xtrabackup for incremental backup , i used –salve-info to incremntal backup for convenience,but i found it always printing >> log scanned up to (0 1172823116) ，so the process of backup countn’t
end , the xtrabackup_suspended file alse exist. when i drop the –salve-info option , the process of backup could be compelte soon.

Comment on The relationship between Innodb Log checkpointing and dirty Buffer pool pages by Shenglin

Shenglin — Wed, 22 Feb 2012 02:58:14 +0000

I had the issue issue with mysql 5.1 version, unfortunately, i can’t tune those parameters besides decreasing innodb_max_dirty_pages_pct to 1 and increase log file size to 3.9G. As I understand,once the checkpoing ago is reaching 7/8 of innodb_log_file_size, flush process use sync instead of async and block all the write operations. How can I verify this? I tried to pstack(Solaris OS), and the insert is waiting for buf_flush_wait_batch_end ().

Comment on How to recover deleted rows from an InnoDB Tablespace by Shenglin

Shenglin — Wed, 22 Feb 2012 02:29:50 +0000

it’s good to have a delay slave if TCO is acceptable

Comment on Setting up XFS on Hardware RAID — the simple edition by Trent Hornibrook

Trent Hornibrook — Tue, 21 Feb 2012 23:47:06 +0000

Erm I mean

856422399 / 128 = 6690799.9921875 – so thats not block aligned. (presume that theres no point unless you preallocate the tablespace to the entire disk?)

Comment on Setting up XFS on Hardware RAID — the simple edition by Trent Hornibrook

Trent Hornibrook — Tue, 21 Feb 2012 23:45:47 +0000

Thoughts on block aligning the end cylinder as well.

414538752 / 128 is not block aligned

So thoughts on using fdisk to set the end cylinder to the biggest block aligned cylinder?

Comment on Can we improve MySQL variable handling ? by Simon J Mudd

Simon J Mudd — Tue, 21 Feb 2012 22:45:14 +0000

Peter, make a bug report (feature request) for this and post the URL. I’m also adjusting some of my posts where I complain about current functionality, as it’s fair to say that if there’s no bug. report it’s possible that the comments just may not get seen. Having the bug report ensures that Oracle see the wanted “feature request”, even if that may take a long time to appear. Then Oracle customers can reference that feature request and say “I want that too”.

Comment on Optimize Your SQL With Percona’s Online Query Advisor! by bob

bob — Tue, 21 Feb 2012 15:02:42 +0000

On the subject of OpenID logins: why login at all?

What the hell is wrong with this captcha system? And why does it keep reselecting “notify me”?

Comment on Announcing MySQL Monitoring Plugins from Percona by Baron Schwartz

Baron Schwartz — Tue, 21 Feb 2012 01:54:16 +0000

In my opinion (backed by a lot of investigation into the causes of server downtime), that plugin monitors everything that causes noisy alerts and /dev/null email filters, and nothing that really indicates or causes actual problems in the server. I’m not ignorant of that plugin; I am building a better mousetrap. Much better.