SoftFIRE Integration Test¶
This project can be used to test the functionality of the experiment-manager and related components.
Python 3.5 or higher.
Git clone the repository and change into the project’s directory. Execute
pip install . for installing the integration test. Afterwards you can start it with the command
Alternatively you can start it by running the project’s softfire-integration-test file with
Copy the integration-test.ini file, which you can find in the project’s etc folder, to /etc/softfire/integration-test.ini. Edit it so that it fits for your testing environment.
This section describes the test phases which are executed by the integration test. Each phase, except for the Create experimenter phase, use the user credentials provided in the configuration file in the experimenter section for accessing the experiment-manager.
First of all the integration test checks if the experiment file exists and scans it for the resources it uses. If something goes wrong in this phase, the remaining phases will not be executed and the test stops.
This phase is optional and is only executed if the create-user property in the configuration file is set to true. It tries to create a new experimenter in the experiment-manager whose username and password are also defined in the configuration file in the experimenter section. This is the only phase in which the admin-username and admin-password properties from the configuration file are used.
In this phase the experiment is uploaded to the experiment-manager. The experiment that shall be uploaded can be specified in the experiment-file property in the configuration file.
This phase deploys the previously uploaded experiment.
In this phase the integration test checks whether all the resources defined in the experiment are running correctly. The exact validation procedure depends on the resource type. At the moment only resources of type NfvResource can be validated. This is done by waiting until the NSRs are in active state and then checking if the floating IP addresses of the NSR are reachable.
The last phase removes the experiment from the experiment-manager.
Adding validators for other NfvResource types¶
Not all of the different resource types can be validated currently. This section describes how you can add the validation functionality for a certain node type to the integration test.
For this you have to create a new module and a new class in the softfire-integration-test project. In particular:
- create a sub-module in the eu.softfire.integrationtest.validators module called as the node type you want to validate (for instance NfvResource for the Nfv Resource)
- create a class in the sub-module you just created called NodeTypeValidator (for instance NfvResourceValidator)
- the class must extend the eu.softfire.integrationtest.validators.AbstractValidator class and implement the abstrac method validate
The abstrac class is:
class AbstractValidator(metaclass=ABCMeta): @abstractmethod def validate(self, resource, resource_id): pass
The Validator class has to inherit from the AbstractValidator class and implement the validate method which gets the resource and the resource ID of the resource to test as an argument. In this method you can implement the validation of the resource. If the validation is successful, the method should simply return, otherwise it should raise an Exception (class of subclass of Exception)with the reason of failure as the Exception argument.
With the get_resource_from_id(resource_id) method you can get the new status of the deployed experiment.
If you need an example look at the already existing NfvResourceValidator class in the validators.py module.