Journal-guided RAID resync
Posted Nov 25, 2009 21:56 UTC (Wed) by neilbrown
Parent article: Journal-guided RAID resync
(1) the correct fix for an inconsistent stripe may not always be clear
For an inconsistency caused by a crash interrupting a write, there is no correctness problem. If there are two credible options (such as a RAID1 with two devices where a block differs) then either option is equally correct. One version might be "newer" than the other, but that doesn't mean it is more "correct".
"correctness" can be an issue when inconsistency is caused by dodgy hardware - e.g. a stuck-bit in a device buffer. But that is a completely different problem.
to post comments)