Re: raid10 redundancy

[Reindl Harald <h.reindl@thelounge.net>]

Am 18.05.21 um 20:51 schrieb Phillip Susi:
> 
> Reindl Harald writes:
> 
>> it's common sense that additional load on drives which have the same
>> history makes a failure one one of them more likely
> 
> "It's common sense" = the logical fallacy of hand waving.  Show me
> statistical evidence.  I have had lightly loaded drives die in under 2
> years and heavily loaded ones last 10 years.  I have replaced failed
> drives in a raid and the other drives with essentially the same wear on
> them lasted for years without another failure.  There does not appear to
> be a strong correlation usage and drive failure.  Certainly not one that
> is so strong that you can claim with a straight face that after the
> first failure, a second one can be expected within X IOPS, and the IOPS
> needed to rebuild the array are a significant fraction of X

do what you want - others like to be better safe then sorry especially 
when there is no longer redundancy and you don't surive any error until 
the rebuild is finished

and yes i replaced last week a 365/24 for years running Seagate *desktop 
drive* in a RAID10 with 50k power up hours but that don't imply that you 
can expect that