On Thu, Sep 18, 2008 at 6:22 AM, Jon Schewe <jpschewe at mtu.net> wrote:
> Yes, that would be great, but we don't have test servers. We're an R&D
> shop so we can handle some mistakes and save money on servers. Best
> practice would be to have a separate set of servers for testing and
> production, but we've decided we can handle the risk. So when we're
> testing admin scripts and configurations we do it on the real server to
> make sure it works right and then commit it to the repository. We do
> much of the testing on our workstations, but at some point it needs to
> be tested on the real server.
>

This does not invalidate my idea. On the svn server, you have temp
checkout. post-commit will checkout in this temp folder, and then run
rsync from it to other servers. No rsync is done w/o commit.

That way, the only problem you might have is if 2 people make changes
on 2 different servers, and one of them commit and the other one not.
That way the work of the second guy will be lost. But still, if you
change a "common" files, ask everybody to make the changes only on one
server, and exclude this server from the rsync post-commit process.


-- 
Svetoslav Milenov (Sunny)

Even the most advanced equipment in the hands of the ignorant is just
a pile of scrap.