Jean-Philippe is correct. You do NOT place static folders in your
release directories. You need to place the UPLOADS folder (or any
other folder like it that has static non-versioned files in it) in the
"shared" directory. You then create a task that makes a new symlink to
the UPLOADS directory.
This flexibility is the heart and soul of Capistrano. You can get it
to do anything that you would need at any stage in your deployment
process. here is a sample task that creates a symlink:
desc "create symlink"
task :db, :roles => [:app] do
send(run_method, "ln -nfs #{shared_path}/UPLOADS
#{current_path}/public/UPLOADS")
end
Hope that puts you on the path to creating your own tasks.
- Ian
On Thu, Mar 19, 2009 at 10:23 AM, Jean-Philippe Moal
<[email protected]> wrote:
>
> Hello,
>
> I didn't know that git could store symlinks ?
> Depending on the deployment recipe I guess there could be problems with
> relative
> symlinks due to the fact that the 'current' deployed directory in default
> capistrano structure is a symlink to a folder located in the releases dir.
>
> The most common (at least I think so) is to create a task (with a callback
> after
> the code deployment; I personally use «after "deploy:symlink"») that will add
> the required symlinks in the public dir that links to real folders located in
> the 'shared' dir, which is already used for pids and logs.
>
> François Montel a écrit :
>> I'm setting up capistrano with an existing project that has folders
>> under /public, where files uploaded from users are stored (ie, /public/
>> photos) that is not under version control. It's not part of the code
>> and it's continually being updated, so I don't want it in my git repo
>> (it's marked with git ignore so it isn't included).
>>
>> The way capistrano seems to work is by checking out a new copy of the
>> git repo every time deploy:update is run. So even if I copy the user
>> data into the folder structure created by capistrano the first time I
>> deploy, the next time I update, a new version will be checked out of
>> the repo and that user data won't be there.
>>
>> What's the preferred way to handle this? I could think of a couple of
>> options:
>>
>> 1. Store the files (ie, photos) in a folder outside of my app
>> structure, and symlink it to public/photos. That symlink would be in
>> my git repo so every fresh checkout would still point to the same user
>> file folder.
>>
>> 2. Maybe after my initial deploy, I shouldn't be using deploy:update
>> more, but my own recipe that just performs git pull and restarts
>> passenger. After all, I don't really want capistrano to perform a
>> whole new checkout every time I update the code (that's time
>> consuming). I just want it to update it with git pull. (Capistrano has
>> nice rollback features, but I can roll back with git if I need to.)
>>
>> Perhaps there are other or better ways to handle this.
>>
>> Thanks for the help!
>>
>> PS. Capistrano 2.3 has a scm: none option, but that seems to be for
>> when the whole app is not under version control; not in the case of
>> subfolders as mentioned above.
>> >
>
>
> >
>
--
----------------------
Ian Sheridan
http://iansheridan.dyndns.org
----------------------
--~--~---------~--~----~------------~-------~--~----~
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at http://groups.google.com/group/capistrano
-~----------~----~----~----~------~----~------~--~---
