.. _setup:

Setup

=====

Setting up the application

--------------------------

::

 paster make-config RhodeCode production.ini

- This will create `production.ini` config inside the directory

  email settings,static files, cache and logging.

::

 paster setup-app production.ini

- This command will create all needed tables and an admin account. 

  When asked for a path You can either use a new location of one with already 

  existing ones. RhodeCode will simply add all new found repositories to 

  it's database. Also make sure You specify correct path to repositories.

- Remember that the given path for mercurial_ repositories must be write 

  accessible for the application. It's very important since RhodeCode web interface

  will work even without such an access but, when trying to do a push it'll 

  eventually fail with permission denied errors. 

- Run 

::

 paster serve production.ini

  127.0.0.1:5000. This ip and port is configurable via the production.ini 

  file  created in previous step

- Use admin account you created to login.

- Default permissions on each repository is read, and owner is admin. So 

  remember to update these if needed.

Setting up Whoosh full text search

----------------------------------

Index for whoosh can be build starting from version 1.1 using paster command

passing repo locations to index, as well as Your config file that stores

whoosh index files locations. There is possible to pass `-f` to the options

to enable full index rebuild. Without that indexing will run always in in

incremental mode.

::

 paster make-index --repo-location=<location for repos> production.ini  

for full index rebuild You can use

::

 pip install ldap-python

ldap-python requires some certain libs on Your system, so before installing it 

check that You have at least `openldap`, and `sasl` libraries.

ldap settings are located in admin->permissions section,

Here's a typical ldap setup::

 Enable ldap  = checked                 #controlls if ldap access is enabled

 Host         = host.domain.org         #acctuall ldap server to connect

 Port         = 389 or 689 for ldaps    #ldap server ports

 Enable LDAPS = unchecked               #enable disable ldaps

 Account      = <account>               #access for ldap server(if required)

 Password     = <password>              #password for ldap server(if required)

 Base DN      = CN=users,DC=host,DC=domain,DC=org

`Account` and `Password` are optional, and used for two-phase ldap 

authentication so those are credentials to access Your ldap, if it doesn't 

support anonymous search/user lookups.

If all data are entered correctly, and `ldap-python` is properly installed

so You can control over permissions even on ldap users. If such user exists 

If You have problems with ldap access and believe You entered correct 

ldap will be saved there.

Nginx virtual host example

--------------------------

Sample config for nginx using proxy::

 server {

    listen          80;

    server_name     hg.myserver.com;

    access_log      /var/log/nginx/rhodecode.access.log;

    error_log       /var/log/nginx/rhodecode.error.log;

    location / {

            root /var/www/rhodecode/rhodecode/public/;

            if (!-f $request_filename){

                proxy_pass      http://127.0.0.1:5000;

            }

            #this is important for https !!!

            proxy_set_header X-Url-Scheme $scheme;

            include         /etc/nginx/proxy.conf;  

    }

 }  

Some extra configuration files and examples can be found here:

http://hg.python-works.com/rhodecode/files/tip/init.d

and also an celeryconfig file can be use from here:

http://hg.python-works.com/rhodecode/files/tip/celeryconfig.py

Troubleshooting

---------------

- missing static files ?

 - make sure either to set the `static_files = true` in the .ini file or

   double check the root path for Your http setup. It should point to 

   for example:

   /home/my-virtual-python/lib/python2.6/site-packages/rhodecode/public

- can't install celery/rabbitmq

 - don't worry RhodeCode works without them too. No extra setup required

- long lasting push timeouts ?

 - make sure You set a longer timeouts in Your proxy/fcgi settings, timeouts

- large pushes timeouts ?

 - make sure You set a proper max_body_size for the http server

.. _virtualenv: http://pypi.python.org/pypi/virtualenv

.. _python: http://www.python.org/

.. _mercurial: http://mercurial.selenic.com/

.. _celery: http://celeryproject.org/

.. _rabbitmq: http://www.rabbitmq.com/

    def get_repo(self, repo_name):

        return self.get(repo_name)

    def get(self, repo_name):

        """

        Get's repository from given name, creates BackendInstance and

        propagates it's data from database with all additional information

        :param repo_name:

        """

        if not HasRepoPermissionAny('repository.read', 'repository.write',

                            'repository.admin')(repo_name, 'get repo check'):

            return

        @cache_region('long_term')

        def _get_repo(repo_name):

            repo_path = os.path.join(self.repos_path, repo_name)

            alias = get_scm(repo_path)[0]

            log.debug('Creating instance of %s repository', alias)

            backend = get_backend(alias)

            #TODO: get the baseui from somewhere for this

            if alias == 'hg':

                #skip hidden web repository

                if repo._get_hidden():

                    return

            else:

                repo = backend(repo_path, create=False)

            dbrepo = self.sa.query(Repository)\

                .options(joinedload(Repository.fork))\

                .options(joinedload(Repository.user))\

                .filter(Repository.repo_name == repo_name)\

                .scalar()

            make_transient(dbrepo)

            repo.dbrepo = dbrepo

            return repo

        invalidate = self._should_invalidate(repo_name)

        if invalidate:

            log.info('invalidating cache for repository %s', repo_name)

            region_invalidate(_get_repo, None, repo_name)

            self._mark_invalidated(invalidate)

        return _get_repo(repo_name)