On Sat, 03 Sep 2011, Luke Kenneth Casson Leighton wrote: > now, you may be thinking "surely, that was just unlucky with two > drives, right?" wrong - the total number of drives used for this RAID1 > mirror was *four* drives [cf: earlier very helpful discussion > involving a script which someone published - thanks! - that detected > partly-complete RAID mirrors]
Same manufacturing lot? Or bought from the same supplier? Have you checked with WD the origin of these drives? These are very important things to check for when doing root-cause analysis on multiple HD failures. I've seen my share of bad HDD manufacturing lots _and_ bad HDD *supplier batches*. The first one is caused by component or manufacturing defects. The second one by negligent handling during storage or transportation, and it is way too common. I am not saying WD Green HDDs are good, but what you describe is way beyond the expected failure rates even for products that aren't engineered or manufactured for high quality in the first place. > according to the mdadm mismatch count as a standard heuristic / Don't trust mdadm mismatch count to gague HDD quality. It is supposed to detect array corruption, not sector errors. Array corruption is a lot more likely to be caused by kernel bugs, operator error, or bad RAM on any modern system where the HBA-HDD link is protected against trivial corruption, than by other hardware issues. Actually, I have this hunch md/mismatch_cnt actually doesn't work half right, but I've not had the time to actually validate it. > now, apparently, what Western Digital do is they test new drives > thoroughly, and if they pass with flying colours, they are labelled > "black" and sold for more money. if they fail, then they're > "re-programmed" to run a bit slower, thus making less noise, use less > power, and can therefore justify being sold with a "green" label. That can actually be checked, at the price of destroying a Black and a Green WD HDD, and wasting a weekend. It is not even very difficult to do. Do you have any references to someone who did that checking? > by contrast, hitachi's 1.5tb drives which were £70 each off of ebuyer > (instead of £55 for the WD Elements including the external USB case) > run consistently at a full FOUR degrees centigrade lower temperature, > even when the WD Green Drive was removed from its USB case and placed > into the exact same server in which the hitachi drive was present. IMHO Hitachi GST is the only maker of 2.5" or 3.5" *SATA* HDDs worth buying ATM. It may or may not last, as WD is buying Hitach GST. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110904203100.ga26...@khazad-dum.debian.net
