[Savannah-hackers-public] Re: [gnu.org #494104] colonialone: 2nd disk of /dev/md2 dead + filesystem errors on newly created LV

[Sylvain Beucler]

Hi,

On Wed, Oct 21, 2009 at 11:49:30AM -0400, Daniel Clark via RT wrote:
> > [beuc - Wed Oct 21 03:17:03 2009]:
> > - Secondly I got a filesystem error on a LV disk I had just created
> >   (on vg_1.0tb, on md1), which was then remounted read-only.  It's
> >   pretty strange, so I wonder if you have anything to say / suggest
> >   about this.
> 
> Maybe we should just stop using the 1TB disks, which are a bit older
> than the 1.5TB disks, and which we do not use in other places.
> 
> I think colonialone now has far more disk than it needs, so unless there
> are objections I'll just bring colonialone down and swap the 2 1TB
> drives in the computer with 2 1.5TB drives in the external enclosure,
> and then savannah-hackers can migrate everything off of the 1TB disks,
> and I'll just remove the external enclosure during the next GNAPs visit,
> which will remove a possible point of failure. 
> 
> Also since there is now so much more disk space than you need, I would
> highly suggest doing quad redundancy in RAID1, as there is often a delay
> of several days before we can get out to GNAPS and replace disks, and
> also disks tend to fail more often when being used to reconstruct a
> broken array. FSF is using triple redundancy RAID1 in all new systems at
> the moment, but will be moving to quad (sort of; 2 clustered RAID1s) soon.
> 
> Please tell me if my impression that a total of 1.5GB of drive space for
> the new savannah is not enough. The current savannah seems to use less
> than 200GB. 

- Now that the cause of the failure is known to be a software failure,
  do we forget about this, or still pursue the plan to remove 1.0TB
  disks that are used nowhere else at the FSF?

- If not, do you recommend moving everything (but '/') on the 1.5TB
  disks?

- About UUIDs, everything in fstab in using mdX, which I'd rather not
  mess with.

-- 
Sylvain