Re: [Savannah-hackers-public] Re: [gnu.org #649274] colonialone ailing

[Sylvain Beucler]

On Tue, Dec 21, 2010 at 11:01:41PM -0500, Bernie Innocenti wrote:
> On Tue, 2010-12-21 at 18:39 -0500, Bernie Innocenti wrote:
> > On Tue, 2010-12-21 at 16:21 -0500, Michael J. Flickinger via RT wrote:
> > > Bernie has started with the upgrade and it's being coordinated in 
> > > #savannah.
> > > 
> > > I setup the downtime page and disabled the savannah-backup cron for the 
> > > duration of the maintenance.
> > 
> > The downtime is projected to last another 2-3 hours.
> > 
> > 
> > We want to ensure that all md's are completely synced before putting the
> > thing back in production. This time we partitioned the disks into
> > smaller raid1 arrays that can be resynced faster:
> > 
> >   http://codewiz.org/pub/savannah
> > 
> > So far, the hardware has been performing well under the considerable I/O
> > load of syncing + copying over data from the old LVs to the new ones. In
> > case we see any problem, we're still in time to back off and return to
> > the original situation.
> 
> Savannah is back up, running off the two new drives. A third one is
> currently resyncing. The forth still contains the old volumes, as an
> extra precaution in case something goes terribly wrong.

Great!

Good that you could kill 2 birds with one stone.

Re tune2fs: I'm not sure it's a good idea to prevent Xen from fsck'ing
the disks, nor about dropping superuser-reserved 5% which is usually
handy when the system needs to mail about disk shortage.

I also changed /root/remote_backup.sh to back up vg_savannah, and not
VG0 anymore - it was still backing up the old data.

-- 
Sylvain