The edge-api project is an API server for fleet edge management capabilities. The API server will provide Restful web services.
Below you can see where the edge-api application sits in respect to the interaction with the user and the device at the edge to be managed.
┌──────────┐
│ USER │
└────┬─────┘
│
│
│
┌──────────────────────────────────────────────┼─────────────────────────────────────────────────────┐
│ │ │
│ cloud.redhat.com ┌────────▼──────────┐ │
│ │ 3-Scale │ │
│ │ API Management │ │
│ └────────┬──────────┘ │
│ │ │
│ │ │
│ │ │
│ ┌───────▼────────┐ │
│ │ edge-api │ │
│ │ application │ │
│ └───────┬────────┘ │
│ │ │
│ work requests │ playbook run results │
│ (playbooks) │ (ansible runner events) │
│ ┌─────────▼───────────┐ │
│ ┌───────────────────────┤ playbook-dispatcher │◄──────────────────────────────┐ │
│ │ │ application │ │ │
│ │ └─────────────────────┘ │ │
│ │ │ │
│ ┌────────▼────────┐ ┌─────┴────┐ │
│ │ cloud-connector │ │ ingress │ │
│ │ service │ │ service │ │
│ └────────┬────────┘ └─────▲────┘ │
│ │ │ │
└────────────┼─────────────────────────────────────────────────────────────────────────────┼─────────┘
│ │
┌────▼─────┐ │
│ MQTT │ │
│ BROKER │ │
└────┬─────┘ uploads │
│ (http) │
│ ┌───────────────────────────┐ │
│ │ │ │
│ │ Connected Host │ │
signals │ │ ┌───────────────────┐ │ │
(mqtt) │ │ │ RHC Client │ │ │
│ │ │ ┌───────────────┐ │ │ │
└────────────────────┼──►│ │playbook-worker│ ├───┼────────────────────────────┘
│ │ └────┬────▲─────┘ │ │
│ │ │ │ │ │
│ └──────┼────┼───────┘ │
│ │ │ │
│ ┌─▼────┴──┐ │
│ │ Ansible │ │
│ │ Runner │ │
│ └─────────┘ │
│ │
└───────────────────────────┘
Development of this project utilizes several tools listed below:
Git is a free and open source distributed version control system designed to handle everything from small to very large projects with speed and efficiency. You can install Git on your system if it's not already available using the following documentation.
Golang is the development code utilized by the edge-api application. You can get setup to develop with Golang by following the install documentation. The dependencies are handled by Go modules and specified in the go.mod file.
Python is only necessary to support the usage of Bonfire, which is used for deployment and testing of the edge-api application. It is recommended to use Python 3.6 with this project. While you may use the Python included with your Operating System you may also find tools like pyenv to be useful for maintaining multiple Python versions. Currently, the development dependencies are obtained using pipenv. Pipenv creates a Python virtual environment with the dependencies specified in the Pipfile and the virtual environment created is unique to this project, allowing different projects to have different dependencies.
Minikube provides a local single node Kubernetes cluster for development purposes. You can find setup information for minikube in the following Get Started! docs. Before starting your cluster you will need to make several configuration updates noted in the Clowder documentation.
Clowder is a kubernetes operator designed to make it easy to deploy applications running on the cloud.redhat.com platform in production, testing and local development environments. This operator normalizes how applications are configured with common interactions, from database to message queue and topics, to object storage. Clowder also helps define consistent mechanisms for driving integration tests with noted application dependencies and Job Invocations. Getting started with Clowder is quite simple using a single command to deploy the operator.
Bonfire is a CLI tool used to deploy ephemeral environments for testing cloud.redhat.com applications. bonfire interacts with a local configuration file to obtain applications' OpenShift templates, process them, and deploy them. There is a Pipfile in this repository that specifies bonfire as a dependency, and if you run the installation steps above, you will have it installed on your virtual environment.
Podman / Docker are used to build a container for edge-api that will run in Kubernetes / Red Hat OpenShift. Get started with Podman following this installation document. Get started with Docker following this installation document.
OpenShift CLI is used by Bonfire because of its templating capabitilies to generate the files that will be deployed to your local Kubernetes cluster. Follow the instructions on Installing the OpenShift CLI to install it in your machine.
For these steps, only git and Docker are necessary from the information above. Keep in mind that you might need to run inside of a Kubernetes cluster if you want to test more complex use-case scenarios.
-
Clone the project.
git clone [email protected]:RedHatInsights/edge-api.git
-
Change directories to the project.
cd edge-api -
Run the migrations to create the database schema (this will download dependencies).
go run cmd/migrate/main.go
-
Run the project in debug mode. Debug mode allows unauthenticated calls and it's essential for local development.
DEBUG=true go run main.go
-
Open another terminal and make sure you can reach the API.
curl -v http://localhost:3000/
If you find an error message saying that you don't have gcc installed, install it.
Keep in mind that if you change the models you might need to run the migrations again. By default, Edge API will run with a sqllite database that will be created in the first run. If you want to use your own postgresql container (which is particularly good for corner cases on migrations and queries), you can do this by:
-
Run the dabatase (example with podman)
podman run -d --name postgresql_database -e POSTGRESQL_USER=user -e POSTGRESQL_PASSWORD=pass -e POSTGRESQL_DATABASE=db -p 5432:5432 rhel8/postgresql-10
-
Add some environment variables
DATABASE=pgsql PGSQL_USER=user PGSQL_PASSWORD=pass PGSQL_HOSTNAME=127.0.0.1 PGSQL_PORT=5432 PGSQL_DATABASE=db
If you prefer to run the edge-api using containers, then you can use the following steps.You should have podman or docker already installed in your machine and a valid account account on quay.io
-
Clone the project.
git clone [email protected]:RedHatInsights/edge-api.git
-
Create an authentication json file, replacing the word token with the generated password provided by quay.io
${HOME}/.config/containers/auth.json "auths": { "quay.io": { "auth": "token..." } } }
-
Login on quay.io Please note that you can use either podman or docker for the example below.
podman login registry.redhat.io
to validate:
podman login --get-login registry.redhat.io
-
Create a new folder outside the cloned edge-api repository where you can store all database artifacts. This will allow you to keep a database that can be reused multiple times. For example, you might want to keep your database under a special backup folder inside your $HOME directory
mkdir $HOME/db_backups -
Change into the directory where you've cloned the edge-api repository
cd edge-api -
Launch a containerized instance of a PostgreSQL database Please note that you can use either podman or docker for the example below
podman run --detach --env POSTGRESQL_DATABASE='edge' --env POSTGRESQL_PASSWORD=pass --env POSTGRESQL_USER=user --name podmandb --publish 5432:5432 --pull=always --volume PATH_TO_YOUR_DATABASE_BACK_FOLDER:/var/lib/pgsql/data:Z registry.redhat.io/rhel8/postgresql-12:latestThe example above uses PostgreSQL 12, but you can try a newer version. The important part is to remember what values you use for all the environment variables, such as POSTGRESQL_DATABASE, POSTGRESQL_PASSWORD, and POSTGRESQL_USER, as they will be required later on You can also use environment variables instead of passing them inline
-
Execute the project migrations
podman run --rm -ti -p 3000:3000 -v $(pwd):/edge-api:Z --env DATABASE=pgsql --env PGSQL_DATABASE=edge --env PGSQL_HOSTNAME=YOUR_IP --env PGSQL_PASSWORD=pass --env PGSQL_PORT=5432 --env PGSQL_USER=user --name edge-api --pull=always quay.io/cloudservices/edge-api:latest /usr/bin/edge-api-migrateYou can test that your containerized database is working by using the following command
psql --host localhost --user user edge
If you don't have the psql command available locally, you can use the containerized instance itself to check the database:
docker exec -it postgres --host localhost --user user edgeType \c edge to connect to the edge database and \dt to list all available tables. Type \q to exit.
Note: You won't have any tables until you have run the migration scripts, as explained further down.
-
Start up the edge-api service
podman run --rm -ti -p 3000:3000 -v $(pwd):/edge-api:Z --env DATABASE=pgsql --env PGSQL_DATABASE=edge --env PGSQL_HOSTNAME=YOUR_IP --env PGSQL_PASSWORD=pass --env PGSQL_PORT=5432 --env PGSQL_USER=user IMAGEBUILDERURL=imagebuilder_url--env INVENTORYURL=inventory_url --env HTTP_PROXY=proxy --env HTTPS_PROXY=proxy --name edge-api --pull=always quay.io/cloudservices/edge-api:latestYou can also use environment variables instead of passing them inline The values for the DB should be the same as you created on DB setup (step 6)
-
In another terminal or tab, test your local environment get
curl -v http://localhost:3000/
curl post
curl --request POST --url localhost:3000/api/edge/v1/device-groups/ --header 'Content-Type: application/json' --data '{"Account": "0000000","Name":"test", "Type":"static"}'
docs
localhost:3000/docs
Following the information above you should have Docker or Podman, a minikube cluster running with Clowder installed, and a Python environment with bonfire installed. Now move on to running the edge-api application.
-
Clone the project.
git clone [email protected]:RedHatInsights/edge-api.git
-
Change directories to the project.
cd edge-api -
Setup your Python virtual environment.
pipenv install --dev
-
Enter the Python virtual environment to enable access to Bonfire.
pipenv shell
-
Setup access to the Docker enviroment within minikube, so you can build images directly to the cluster's registry.
eval $(minikube -p minikube docker-env)
-
Build the container image.
make build
-
Create Bonfire configuration. To deploy from your local repository run the following:
make bonfire-config-local
The above command will create a file named
default_config.yamlpointing to your local repository. At times you may need to update the branch which is referred to with therefparameter (defaults tomain).Bonfire can also deploy from GitHub. Running the following command will setup the GitHub based configuration:
make bonfire-config-github
-
Setup test namespace for deployment.
make create-ns NAMESPACE=test
-
Deploy a Clowder environment (ClowdEnviroment) to the namespace with bonfire.
make deploy-env NAMESPACE=test
-
Deploy the application to the namespace.
make deploy-app NAMESPACE=test
Now the application should be running. You can test this by port-forwarding the app in one terminal and running a curl command in another as follows:
Terminal 1
kubectl -n test port-forward service/edge-api-service 8000:8000Terminal 2
curl -v http://localhost:8000/You should get a 200 response back.
Now you can build and deploy the application.
Once a code change has been performed you can rebuild the container in the minikube registry:
make buildThen scale down and up the deployment to pick up the image change:
make restart-app NAMESPACE=testThis project makes use of Golang's built in lint and vet capabilites. You can run these with the following commands:
lint:
make lintvet:
make vetGolang also provides a unit test infrastructure test. Run unit tests with the following command:
make testkin-openapi is a tool that helps us handle docs in the format of Open API spec. Sadly, it does not generate docs automagically. We have a script that generates the docs and you have to add your model there to be picked by the code generation.
The openapi3gen package generates the docs for the models in the project and we have to update the routes by hand on the path.yml file. Our generation script adds the routes that you wrote by hand, creating an openapi.json and a openapi.yaml.
You have to commit and create a pull-request to update the docs.
To run the comand that generates the docs, you can use:
make generate-docsThe API will serve the docs under a /docs endpoint.
The tests in this repository are meant to be unit tests. Because of how Go works, in order to mock objects properly you can't create them inside of your function. You can instead receive objects that are supposed to be mocked through function parameters, context, or variables inside of the same struct.
We are using gomock to generate mocks for our unit tests. The mocks are living inside of a package under the real implementation, prefixed by mock_. An example is the package mock_services under pkg/services.
In order to genenerate a mock for a particular file, you can run:
go install github.com/golang/mock/mockgen@latest
mockgen -source=pkg/filename.go -destination=pkg/mock_pkg/mock_filename.go
For example, to create/update mocks for ImageService, we can run:
mockgen -source=pkg/services/images.go -destination=pkg/services/mock_services/images.go