Jeff,
A restore isn't meant to deal with the situation you describe. The user
(application group) is responsible for keeping the information about
their files in order. So if they accidentally remove an entry they can
bring it in again with the usual means. Of course you stay with the
dangling physical files. An approach could be that you simply remove
physical files without an RLS entry, something that in the past is
suggested too I think. But I'm not aware of the current thinking on the
policies for this.
Jules
> -----Original Message-----
> From: LHC Computer Grid - Rollout
> [mailto:[log in to unmask]] On Behalf Of Jeff Templon
> Sent: dinsdag 18 januari 2005 11:36
> To: [log in to unmask]
> Subject: Re: [LCG-ROLLOUT] [ATLAS-LCG] Disk failure at Prague
>
> Hi Jules, *,
>
> I think what Stephen meant was that a user, while trying to update the
> RLS database (not the data file), might accidentally delete entries or
> modify them. The question then is "how long will it be before someone
> realizes that some entries have been deleted from the RLS"?
> Then if you
> try to restore from a RLS backup, you have to resolve all the problems
> between the current (C) and backup (B) states:
>
> which of the entries present in B but not in C correspond to
> accidentally-deleted entries, and which correspond to intentionally
> (properly) deleted ones?
>
> JT
>
> On Tue, 2005-01-18 at 09:18, Jules Wolfrat wrote:
> >
> > > -----Original Message-----
> > > From: LHC Computer Grid - Rollout
> > > [mailto:[log in to unmask]] On Behalf Of
> Burke, S (Stephen)
> > > Sent: maandag 17 januari 2005 17:46
> > > To: [log in to unmask]
> > > Subject: Re: [LCG-ROLLOUT] [ATLAS-LCG] Disk failure at Prague
> > >
> > > LHC Computer Grid - Rollout
> > > > [mailto:[log in to unmask]] On Behalf Of Jules
> > > Wolfrat said:
> > > > The VO doesn't have to notice, the sysadmins are supposed to
> > > > detect that
> > > > the RLS is broken (e.g. a disk crash) and have to restore,
> > > at least at
> > > > our site. And we keep backups at least for several weeks.
> > >
> > > I wasn't thinking of crashes so much as deliberate or
> > > accidental corruption
> > > by a user/hacker, which might well go un-noticed until
> > > someone tries to
> > > access one of the missing files.
> >
> > That's true, but you must make a distinction between the
> database that
> > RLS is and the physical file itself. The file itself can be
> destroyed
> > while the RLS itself still thinks that the file is present. What you
> > need is some mechanism to check the integrity of the RLS,
> e.g. run every
> > night a consistency check on the RLS!
> >
> > Jules
> > >
> > > Stephen
> > >
>
|