My apoligies for not being able to turn up to the meeting, the laptop
on which I have tamed evo had an an accident involving a leaky bag and
lancaster weather and I had no luck getting my home desktop machine to
play nice with evo.
I would like to discuss our problems at a later storage meeting, not
so much the details (I'll be blogging about those today or tomorrow)
but some of the more nebulous operational questions- like how much
downtime is data worth? Okay, that does largely depend on several
factors such as the specifics of the site and the data that might be
lost-but roughly can be boiled down to should data recovery be on a
"best effort" basis or should you go "above and beyond". There's a
third possible point of view, "bah user data, loser data" where you
just get yourself up and running as fast as possible, which again
might be applicable to some sites- but I think that would be the
minority case.
Linked to the above question, if you're in the odd state like I'm in
where you have the data in some form but can't get to it, how long
would be advisible to stay in this data limbo for? I know it suffers
from the nebulousness as my previous enquiry but some vague guidelines
would be useful. I've had no "I can't get file X" tickets from any VOs
yet, and the dcache wizards are still conjuring up a solution that
might save some/all of the lost data (considering the season and the
fact that I'm a naturally lower priority then the big tier 1s this may
take a while). So I don't want to rush things- and if/when I do have
to clean up and publish a list of "lost" data I can purge the files
with one of my cleanup scripts and pull a list of the lost files for
the various VOs from my billing database (which is thankfully intact
on the srm host).
Yikes, that was a tad long-winded, but those were the questions I
wanted to put to the storage meeting today.
cheers,
Matt
On 12/12/2007, Greig Alan Cowan <[log in to unmask]> wrote:
> Hi all,
>
> In addition to the agenda, we should talk about Lancaster's recent
> problems. DPM on SL4 has also been generating a lot if interest.
>
> A report on the Tier-1 dCache upgrade would also be good.
>
> Cheers,
> Greig
>
>
> On 11/12/07 11:21, Jensen, J (Jens) wrote:
> > http://indico.cern.ch/conferenceDisplay.py?confId=25237
> >
> > EVO, usual location, usual password.
> >
> > Cheers
> > --jens
>
|