The network policy viewer visualizes the pod network. It is far from complete, but basic isolation rules can be represented in JSON, YAML or dot (Graphviz).
In this example, the names of the namespaces match their respective network policies, the exception being the global
namespace (which has none) and ingress-isolated-whitelist
(which has two).
The policies isolated
, egress-isolated
, ingress-isolated
each apply to the namespace as a whole.
ingress-isolated-whitelist
whitelists httpd-bob
, which is why httpd-bob
can be reached from httpd-alice
and the generic httpd pod in the namespace.
Install the helm chart defined in the folder chart:
$ make -C chart install
Point your browser to the URL given in values.yaml
(e.g. http://minikube.info/
):
The available endpoints are:
Endpoint | Description |
---|---|
/ | Show graph |
/health | Health endpoint |
/api/v1/metrics | Metrics endpoint |
The build steps are the following:
$ go mod download
$ go get
$ go vet
$ go test -v
$ go build -o k8s-network-policy-viewer .
make build
will run these steps in a two-stage docker build process.
Alternatively, you can use the default image k8s-network-policy-viewer. This is also the image referenced in the helm chart.
To build the sample data, run:
$ make -C testdata init
$ make -C testdata create
The application is intended for in-cluster use -- in which case it fetches the required API resources from the cluster -- but you can supply arbitrary input by piping the output of kubectl get pod,namespace,networkpolicy --all-namespaces -o json
to it. The application accepts both JSON and YAML, but you may wish to work with JSON so you can filter the input with jq
.