Skip to content

Configuring Junction

Junction enables testing to validate behavior with the same configuration that eventually gets used dynamically. In this last part of the getting started guide, we walk through how the same configuration can be used in three different ways.

  • Static configuration for unit testing
  • Static configuration with dynamic IP's for Integration Testing
  • Dynamic configuration with ezbake

Defining our config

To start, we need to define a configuration. Here we have a route on the hostname jct-simple-app.default.svc.cluster.local, which sends all requests matching the path /v2/user to a test service called jct-simple-app-v2, putting in place a retry policy as well:

import junction
import junction.config
import typing

my_service = {"type": "kube", "name": "jct-simple-app", "namespace": "default"}
my_test_service = {"type": "kube", "name": "jct-simple-app-v2", "namespace": "default"}

retry_policy: junction.config.RouteRetry = {
    "attempts": 3,
    "backoff": 0.5,
    "codes": [500, 503],
}

routes: typing.List[junction.config.Route] = [
    {
        "id": "jct-simple-app-routes",
        "hostnames": ["jct-simple-app.default.svc.cluster.local"],
        "rules": [
            {
                "backends": [{**my_test_service, "port": 8008}],
                "retry": retry_policy,
                "matches": [{"path": {"value": "/v2/users"}}],
            },
            {
                "backends": [{**my_service, "port": 8008}],
                "retry": retry_policy,
            },
        ],
    },
]

Static configuration for unit testing

The first step in testing a configuration is unit tests. Junction provides the check_route method, which lets you test how a specific request will get processed by it's rule:

# assert that requests with no path go to the cool service like normal
(_, _, matched_backend) = junction.check_route(
    routes, "GET", "http://jct-simple-app.default.svc.cluster.local", {}
)
assert matched_backend["name"] == "jct-simple-app"

# assert that requests to /v2/users go to the cool-test service
(_, _, matched_backend) = junction.check_route(
    routes, "GET", "http://jct-simple-app.default.svc.cluster.local/v2/users", {}
)
assert matched_backend["name"] == "jct-simple-app-v2"

Static configuration for pre-deployment testing

Before we roll out the configuration dynamically, we probably want to see it work in a real HTTP client. To allow this mode, all clients can be configured with static routes and backends, and use those rather than what comes back from the control plane.

Then: 

import junction.requests as requests

session = junction.requests.Session(
    static_routes=routes
)
r1 = session.get("http://jct-simple-app.default.svc.cluster.local")
r2 = session.get("http://jct-simple-app.default.svc.cluster.local/v2/users")
# both go to expected service
print(r1.text)
print(r2.text)

Dynamic configuration with ezbake

The final step is deploying the configuration to the control plane feeding all clients in the cluster. Junction provides integrations with EZBake to make this simple. EZBake uses the Gateway API HTTPRoute to specify routes, and so junction provides a method that allows configuration to be dumped to a file, that can either be directly executed with kubectl apply -f, or put in whatever GitOps mechanism you use to roll out configuration. 

for route in routes:
    print("---")
    print(junction.dump_kube_route(route=route, namespace="default"))