On 21/01/2011 15:55, Bill Moran wrote:
> On the "no" side, doing this kind of thing is always complex. We have a
> slew of other, very specialized scripts that do things like convert a
> production database to a development database by sanitizing sensitive
> data, or automatically deploy new database or upgrades to either our
> production, development, or lab environments. Build tools like phing,
> make, ant, etc make this kind of thing easier to create, but it's still
> a lot of work because each situation is special.
Yes, I've also made a set of migration and sanitizing scripts so that
aspect is covered. I was only wondering if there is some way I'm missing
that would solve this one step more elegantly.
> So, my overall answer is that you're probably on the right track, you
> probably don't realize how much work is involved to get this really
> working well, and I would change just a few things about your approach
> based on the information you've provided so far. The requirement to
> use an unprivileged user to restore is going to make the tools you
> use to prepare the input files more important, but that's not insurmountable
> if you keep control over them (i.e. use schema files and a tool for
> normalizing them and do the data dump separately)
Thanks, I'll consider that approach.