Re: Repository recovery and archiving

[Jesus M. Salvo Jr.]

In short, I need the transaction capabilities of a database ( going back
to "cvs tag" needing to be "all" or "none," not "some" in the event of
failure during the operation of "cvs tag" ... since tagging multiple
files are not atomic ) in a source control.

What I am looking at now is Oracle's iFS ( Internet File System ), which
is basically a file system in a database ( that is accessible via HTTPS,
LDAP, SMTP, IMAP, etc. ) that has version control capabilities. Not sure
yet if they have the "tagging" capabilities though.

At least with this, I have last night's backup plus the archive logs of
today's transaction so I can restore from last night's backup and
"replay" today's transactions via the archive logs to bring the
repository up to the last time the archive log was updated.


John



Rob Helmer wrote:
> 
> Ah, ok, well you can check files out by date, but if that
> revision was tagged after that date then you're going to get
> tags.
> 
> Doing incremental backups with Amanda, dump, by scripting tar
> or any other backup system is probably the way to go.
> 
> Can you explain why you need to be able to do this? Perhaps
> you could do a backup right before a build and right before
> a release, and assign your build or release tag after the
> backup finishes?
> 
> > I understand that, and the problem with that is you are restricted to at
> > least restoring from last night's backup.
> >
> > I would like to be able to restore the state ( consistent and all ) of
> > the CVS repository to what it was an hour ago.
> >
> > And the restore has to be consistent, in that if someone made a 'cvs
> > tag' on the entire module during that hour:minute:second, the restore
> > should either not have this tag completely or have this tag on all files
> > that would have been tagged ... not some were tagged and some were not
> > tagged.
> >
> >
> > John
> >
> >
> > Rob Helmer wrote:
> > >
> > > Hi Jesus,
> > >
> > > Since they are just RCS files on the backend, you can just
> > > create a tarball or use a common backup system ( I used to use
> > > Arkeia for GNU/Linux, I was able to backup and restore the
> > > archive without incident ).
> > >
> > > I use tar to backup straight to tape on one of my servers, I've
> > > never had a problem restoring.
> > >
> > > Please correct me if I'm missing something important :)
> > >
> > > As an aside, I wouldn't consider an HA architecture to be
> > > a substitute for a consistent backup policy.. there's never
> > > a reason not to archive to a decent storage medium on a
> > > regular basis.
> > >
> > > HTH,
> > > Rob Helmer
> > > Namodn
> > >
> > > On Sat, Mar 10, 2001 at 04:27:43AM +0000, Jesus M. Salvo Jr. wrote:
> > > >
> > > > I am thinking of using CVS as a repository for an internal process for a
> > > > client. The files are a combination of text and binary files.
> > > >
> > > > However, they are under a penalty clause such that if they dont deliver
> > > > the data within the specified time in each day, they get penalised.
> > > > Therefore, restoring a CVS repository from yesterday's backup when
> > > > something went wrong at 1:00pm is the last option. ( Yes, they can
> > > > probably have some HA architecture in place, but they dont and is
> > > > probably out of reach for them ).
> > > >
> > > > So I was wondering if there are documentation / tools for backup and
> > > > recovery of CVS repositories. Meaning, I would like to recover from the
> > > > CVS repositories' consistent state at least an hour ago.
> > > >
> > > > For achiving, some of the files that are about 1 year old need not be in
> > > > the CVS repository, so there must be some option to archive ( place on
> > > > separate storage apart from the actual CVS repository ) those files.
> > > >
> > > > Has anyone done these?
> > > > Are these possible?
> > > >
> > > >
> > > > The more I think of it, it seems that I need some of the features of a
> > > > database but specifically for dealing with files. That is probably my
> > > > second option if I cannot address these issues to the client.
> > > >
> > > > Thanks,
> > > >
> > > > John
> > > >
> > > >
> > > > --
> > > > Homepage: http://homepages.tig.com.au/~jmsalvo/
> > > > Public Key:
> > > > http://pgp5.ai.mit.edu:11371/pks/lookup?op=get&search=0x51F47D34
> > > >
> > > > _______________________________________________
> > > > Info-cvs mailing list
> > > > address@hidden
> > > > http://mail.gnu.org/mailman/listinfo/info-cvs
> > > >
> >
> > --
> > Homepage: http://homepages.tig.com.au/~jmsalvo/
> > Public Key:
> > http://pgp5.ai.mit.edu:11371/pks/lookup?op=get&search=0x51F47D34
> >
> 
> _______________________________________________
> Info-cvs mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/info-cvs

-- 
Homepage: http://homepages.tig.com.au/~jmsalvo/
Public Key:
http://pgp5.ai.mit.edu:11371/pks/lookup?op=get&search=0x51F47D34