2009-05-07 - production
19:47	<domas>	bad disk on db19 actually made I/Os time out, thus corrupting relay logs, reset slave seems to have helped.	[production]