I've just taken over a few PHP app which had no deployment process at
all (and no SCM), so my first step (towards capistrano) has been an
rsync solution. However, not the sort of solution your client has.
My release process is designed to eventually be taken over by
capistrano, so I have the same structure as it uses with the following
in /srv/ao:
current -> releases/2010-01-06
releases
shared
My release process (which is manual right now) is this:
ssh to apa...@hostname
cd ~/src/ao
git pull origin tag _tagname_
cd /srv/ao
cp -al current/ releases/_tagname_
rsync -ruv ~/src/ao/www/ releases/_tagname_
mysql --options dbname < ~/src/ao/sql/up.sql
ln -snf current releases/_tagname_
No need to have any exclusions because I don't do the rsync with the --
delete (yet) - and I don't have any of the user-generated content in
git.
I moved to this in about half an hour from having a Plesk setup with a
single /var/www/vhosts/_hostname_/httpdocs directory. Just created
the new structure, copied the content, updated the DocumentRoot, and
restarted httpd.
The next step, btw, is what I'm doing now which is pulling all the
user-generated content out into the shared directory and having it
symlinked to from within the releases directory so I can use --delete
in my rsync.
HTH.
On Jan 15, 2010, at 9:38 AM, Matt Aimonetti wrote:
Wow, I agree with both Nick and Patrick... something much be wrong
with me ;)
I agree with Patrick that in general you should stick to standard
tools like cap (or vlad). However cap isn't always very efficient
and it might not be able to do something very specific you have to
do. If that's the case, using a simpler solution like a custom git
based solution can be very cool.
However in this very specific case you probably won't the only one
having to deal with deployment, in the future, another contractor
might have to work on the code. If I were you, I'd be kind and not
use some home made deployment solution that only you understand,
unless you want to force the client to depend on you ;)
- Matt
On Fri, Jan 15, 2010 at 8:06 AM, Patrick Crowley
<[email protected]> wrote:
> Anyone do the syncing thing rather than Capistrano? Do you do
anything, maybe with Git, to be ready to roll back and sync the last
known working version of the code?
Unless there is specific technical reason to not use Capistrano
(which I doubt), I'd strongly recommend your client switch over to
Cap. (Vlad is a good option too.)
The roll-your-own-deploy-script-with-rsync-and-ruby approach seems
ripe for failure.
-- Patrick
--
SD Ruby mailing list
[email protected]
http://groups.google.com/group/sdruby
--
SD Ruby mailing list
[email protected]
http://groups.google.com/group/sdruby
--
SD Ruby mailing list
[email protected]
http://groups.google.com/group/sdruby
