[Zope-CMF] Re: Re: [Plone-users] Date based snapshots

[Dylan Jay]

"Paul Winkler" <pw_lists at slinkp.com> wrote in message
news:20040504144747.GA1694 at slinkp.com...
> On Tue, May 04, 2004 at 09:13:11AM +1000, Dylan Jay wrote:
> > "Paul Winkler" <pw_lists at slinkp.com> wrote in message
> > news:20040503145305.GA1684 at slinkp.com...
> > > On Mon, May 03, 2004 at 03:57:43PM +1000, Dylan Jay wrote:
> > > > I have a requirement to be able to view an entire site as it was a
> > certain
> > > > date including pictures etc. One way would be just to copy the site
and
> > > > windback the data.fs. Is there some other way this could be done via
a
> > UI
> > > > sort of like a reverse Vesion object?
> > >
> > > Nothing out of the box.
> > > My first thought would be something like:
> > >
> > > - take daily backups of your database
> > > - use DBTab to mount the backups read-only at different folders
> >
> > I don't think this is going to work as things like portal_catalog and
> > portal_workflow have to viewed at they were too.
>
> I mean something like the following, where all work is
> done in a top-level zope folder called "current" which
> is mounted to its own zodb:
>
> Zope root / current /
>                     plone_1/
>                     plone_2/
>                     ...
>            snapshot_1/
>                       plone_1/
>                       plone_2/
>                       ...
>            snapshot_2
>            ...
>
>
> The snapshot folders would each point to a separate storage.
> The storages could be rotated by a shell script that runs nightly.
> Something like this:
>
> zopectl stop
> N=5            # or however many backups you want to keep.
>
> rm Data.fs.$N  # delete the oldest one.
> for i in `seq $N -1 2`; do     # this is like python's range(N,1,-1)
>     let j=$((i - 1))
>     mv Data.fs.$j Data.fs.$i   # e.g. mv Data.fs.4 Data.fs.5
> done
> cp -f Data.fs Data.fs.1
>
> zopectl start
>
> > You could have a seperate
> > plone site for each folder I guess but that is probably a lot of
storage.
>
> Yes it is. The tradeoff is instant access to all archived material.
> A more complex system could probably be built around incremental
> backups using e.g. repozo.py but the snapshots wouldn't really be
> "live" all the time.

This setup certainly looks not to hard too do and it will allow for viewing
the whole site at certain dates. I can't really see how it will allow diffs
between different versions :(