[ ... ] > It is (euphemism) mysterious that 'rsec/s' is nearly twice as > high as 'wsec/s', and (euphemism) obscure that the RAID set is > also 'sync'ing: Oops I meant "sync"'ed. It is actually slightly obscure because twice as many reads as writes even in degraded mode should not happen, if writes are somewhat naturally aligned. It would be easier to think that MD is rebuilding, but it is not. >> Number Major Minor RaidDevice State >> 0 8 0 0 active sync /dev/sda >> [ ... ] >> 8 8 96 6 active sync /dev/sdg >> 7 0 0 7 removed Perhaps writes are indeed poorly aligned and a way to get this (euphemism) astutely degraded-by-design setup to be less slow would be to greatly increase the stripe cache size, but that might not work, as this reply by NeilB for a a similar previous case hints that things are somewhat complicated: http://www.spinics.net/lists/raid/msg38922.html BAARF.com! :-) [ ... ] -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
