This is an example of a deployment of Sirikata. The scripts in this repository help you setup, build, and run various Sirikata services as well as keep them up to date.
All contents of this repository are available under the BSD license. See the LICENSE file for details.
There's only one driver script to manage deployments, scripts/admin. It manages three components: packages, services, and templates.
Packages -- a build of Sirikata containing binaries you'll use.
Services -- an instance of a binary which you need to build your system, for example one space server or object host. A simple deployment could be made up of the services 'myworld-space' and 'myworld-objects'.
Templates -- descriptions of a set of services which let you.
To get running, you'll first need to setup a package. First initialize the package:
python scripts/admin package init my_package
This doesn't do much, just setting up some space under a data/ directory and placing an empty config.py which you can edit to adjust the default settings. See scripts/admin/packageconfig.py for details. By default it's setup to build from the git repository with a reasonable directory layout. Now you can build the package, which will build and install sirikata within data/packages/my_package
python scripts/admin package build my_package
When it completes you should see that the source code checkout is in data/packages/my_package/build/ and the compiled, installed version is in data/packages/my_package/install/. If you need to in the future, you can modify the code and run build and install commands manually or through the admin script to test new versions of code.
You can also use pre-built binary packages:
python scripts/admin package install my_package http://myhost.com/path/to/sirikata.tar.gz
Most deployments require multiple services, and templates are easier than managing those directly, so we'll talk about them first. You can see an example template configuration in templates/melville. Each template has a top-level template.py which adjusts defaults configuration options for this template. See scripts/admin/templateconfig.py for options.
Then, each subdirectory describes a service template. Note that these do not necessarily map to instantiated services because each might be instantiated multiple times. For example, you may need two space servers using the same configuration and one object host. You only need one service template for the space servers.
Each service template contains a config.py to adjust default settings need one service template for the space servers (see serviceconfig.py for defaults) as well as all data needed to run the service. The melville template's space only contains space.cfg (an easier way to adjust commandline settings than putting them all in one list in config.py). The world template, however, contains a scene file and a script for an object in the scene in addition to a configuration file.
To use the template, we instantiate a copy of it with a prefix to guarantee unique names:
python scripts/admin template init mymelville melville my_package
All the template commands are immediately followed by the prefix, mymelville in this case, for consistency. So the previous command said to instantiate the template 'melville' with the prefix 'mymelville' and use the package 'my_package' as the source of binaries. You should now see the services mymelville-space and mymelville-world under data/services/.
Now to actually run the services:
python scripts/admin template start mymelville
This may take a few moments as the template.py specifies an ordering and timing for services to start up so you can ensure, for example, that the space server is running before the object host starts. Similarly, shutdown the services with
python scripts/admin template stop mymelville
If you want to completely remove the configuration and data associated with the instantiated template, use:
python scripts/admin template destroy mymelville
All the directories that had been created under data/services/ for this template should now be deleted.
If you are creating your own template or debugging a service, you may need to control them directly. Since templates just control collections of services, many of the commands look similar:
python scripts/admin service init service_name package [template/path/]
python scripts/admin service start service_name
python scripts/admin service stop service_name
python scripts/admin service destroy service_name
On command that doesn't apply to templates is for debugging:
python scripts/admin service debug service_name
This works similarly to the start command but adds a debug flag so you will be dropped in a gdb prompt with all the command line flags and working directory setup properly.
You can run services under monit to ensure the recover from crashes or failures. Simply add monit = True to the service configuration. The service and template commands for starting and stopping will place the processes under monit's control if they find that flag enabled.
You can bypass monit when debugging services using the rawstart and rawstop commands, which always execute the final command that starts/stops the service:
python scripts/admin service rawstart service_name
python scripts/admin service rawstop service_name
This is useful if you want to debug the service without having to edit the configuration files.
If you need details on any command, use the help command, e.g.:
python scripts/admin help package install