Comments on: Magic Innodb Recovery self healing http://www.mysqlperformanceblog.com/2007/05/22/magic-innodb-recovery-self-healing/ Everything about MySQL Performance Sat, 21 Nov 2009 05:23:57 -0800 http://wordpress.org/?v=2.8.4 hourly 1 By: Pythian Group Blog » Log Buffer #46: a Carnival of the Vanities for DBAs http://www.mysqlperformanceblog.com/2007/05/22/magic-innodb-recovery-self-healing/comment-page-1/#comment-130836 Pythian Group Blog » Log Buffer #46: a Carnival of the Vanities for DBAs Fri, 25 May 2007 18:40:26 +0000 http://www.mysqlperformanceblog.com/2007/05/22/magic-innodb-recovery-self-healing/#comment-130836 [...] On the MySQL Performance Blog, Peter Zaitsev, another guy with a lot of insight, shows us a little about the ins and outs of dealing with corruption of InnoDB tables. [...] [...] On the MySQL Performance Blog, Peter Zaitsev, another guy with a lot of insight, shows us a little about the ins and outs of dealing with corruption of InnoDB tables. [...]

]]> By: peter http://www.mysqlperformanceblog.com/2007/05/22/magic-innodb-recovery-self-healing/comment-page-1/#comment-129874 peter Tue, 22 May 2007 21:36:41 +0000 http://www.mysqlperformanceblog.com/2007/05/22/magic-innodb-recovery-self-healing/#comment-129874 This situation got interesting development today. Several hours later we have completed ALTER TABLE run for all potentially affected tables and almost going to put box in production I decided to wait until insert buffer merge is completed. This process took about 2 hours and as I was hoping it will complete successfully it crashed. This time it did not crash on applying log records to the database but at later stage as Insert buffer was merged, so corruption probably was in buffer pool. So now we've got to do dump and reload or restore from backup our little 1.3TB database which surely going to be a lot of fun. I wish there would be tools to examine insert-buffer content and possibly discard it. I'd rather have few tables with corrupted indexes which I can rebuild than this. This situation got interesting development today. Several hours later we have completed ALTER TABLE run for all potentially affected tables and almost going to put box in production I decided to wait until insert buffer merge is completed.

This process took about 2 hours and as I was hoping it will complete successfully it crashed.

This time it did not crash on applying log records to the database but at later stage as Insert buffer was merged, so corruption probably was in buffer pool.

So now we’ve got to do dump and reload or restore from backup our little 1.3TB database which surely going to be a lot of fun.

I wish there would be tools to examine insert-buffer content and possibly discard it. I’d rather have few tables with corrupted indexes which I can rebuild than this.

]]>