On Thu, Feb 24, 2005 at 09:10:32AM -0500, Greg Whalin <[EMAIL PROTECTED]> wrote:
Exactly. No ACID database can ensure integerity in such a situation. Postgres, Oracle, or any other transactional DB would have suffered the same fate in these two cases (LiveJournal, Wiki).
FWIW, my understanding of LiveJournal's integrity problem after the power outage involved tables they weren't yet able to migrate off of MyISAM, and getting replication content in all directions. The stuff migrated to InnoDB already came up fine. Their replication setup is a bit complex -- see http://www.danga.com/words/2004_mysqlcon/mysql-slides.pdf for an overview.
(My kit at the same facility was hit too, and recovered correctly, for what it's worth. :-)
At least this mailing list has progressed beyond "Why didn't they have a UPS?", I suppose. :-)
-Rich
They had most of their dbs transitioned to innodb, but even some of those came up corrupted due to write caching being enabled on individual drives. Their myisam tables simply needed indexes rebuilt (which is a pain in the butt and takes forever). Their complete story: http://www.livejournal.com/community/lj_dev/670215.html
Needless to say, between LJ and Wiki, I am fairly paranoid about db corruption now (though our writes are considerably less than either of those two places as we are 99% reads ... our reads are considerably less than either of those sites as well :) )
Greg
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
