xfs_repair of critical volume

[Stan Hoeppner]

Emmanuel Florac put forth on 11/13/2010 2:19 AM:
> Le Fri, 12 Nov 2010 16:14:52 -0600 vous écriviez:
> 
>> This isn't "robustness" Michael.  If anything it's a serious problem.
> 
> I beg to disagree. Would it be better if instead of still having some
> of the data, everything was lost? At what level of accidental
> destruction do you think that the whole data set should be made
> unavailable? 10%? 5? 1? 

You've missed the point of this sub thread discussion, or I did.  He
stated that having the metadata show the files still exist is a positive
thing.  The files are gone.  I stated that this discrepancy is not good
thing.

I believe you are confused, thinking this micro discussion is dealing
with the OP's overall situation.  It is not.  It is dealing strictly
with the issue of the lost set of disks, the files that were on them,
and the fact the metadata says they still exist.  I believe this is due
to the fact that he hasn't run a destructive xfs_repair yet, which I'm
guessing will remove those orphaned metadata entries.

Again, I'm pretty sure you misunderstood exactly what we were talking
about, or I misunderstood what he was talking about, heck, maybe both.
I absolutely was not stating anything akin to throwing the baby out with
the bath water.

-- 
Stan