This demo will guide you through creating two clusters. Each cluster will contain a SPIRE deployment along with the SPIRE Controller Manager. The first cluster (cluster1) will host a greeter server workload. The second cluster (cluster2) will host a greeter client workload. Each cluster is a distinct SPIFFE trust domain.
The greeter server and greeter client communicate with each other over TLS and perform mutual authentication.
To facilitate this cross-cluster authentication, each cluster will be federated with the other by deploying a ClusterFederatedTrustDomain CRD in each cluster that directs that cluster to the SPIFFE Bundle Endpoint for the other cluster.
Additionally, each workload will obtain an X509-SVID from SPIRE. The workload registration will be accomplished by deploying a ClusterSPIFFEID CRD in each cluster that targets the greeter workloads and assigns them the proper ID.
Build the greeter server and client:
$ (cd greeter; make docker-build)
Pull the requisite images:
$ echo ghcr.io/spiffe/spire-server:1.7.0 \
ghcr.io/spiffe/spire-agent:1.7.0 \
ghcr.io/spiffe/spiffe-csi-driver:0.2.3 \
ghcr.io/spiffe/spire-controller-manager:nightly \
| xargs -n1 docker pull
Start up cluster1 and load the requisite images:
$ ./cluster1 kind create cluster
$ echo \
ghcr.io/spiffe/spire-server:1.7.0 \
ghcr.io/spiffe/spire-agent:1.7.0 \
ghcr.io/spiffe/spiffe-csi-driver:0.2.3 \
ghcr.io/spiffe/spire-controller-manager:nightly \
greeter-server:demo \
| xargs -n1 ./cluster1 kind load docker-image
Start up cluster 2 and load the requisite images:
$ ./cluster2 kind create cluster
$ echo \
ghcr.io/spiffe/spire-server:1.7.0 \
ghcr.io/spiffe/spire-agent:1.7.0 \
ghcr.io/spiffe/spiffe-csi-driver:nightly \
ghcr.io/spiffe/spire-controller-manager:nightly \
greeter-client:demo \
| xargs -n1 ./cluster2 kind load docker-image
Deploy SPIRE components in cluster1:
$ ./cluster1 kubectl apply -k config/cluster1
Deploy SPIRE components in cluster2:
$ ./cluster2 kubectl apply -k config/cluster2
Federate cluster1 with cluster2:
$ ./cluster1 scripts/make-cluster-federated-trust-domain.sh | \
./cluster2 kubectl apply -f -
Federate cluster2 with cluster1:
$ ./cluster2 scripts/make-cluster-federated-trust-domain.sh | \
./cluster1 kubectl apply -f -
Deploy the greeter server in cluster1:
$ ./cluster1 kubectl apply -k config/cluster1/greeter-server
Configure the greeter client with the address of the server:
$ ./cluster2 kubectl apply -f - <<EOF
apiVersion: v1
kind: ConfigMap
metadata:
name: greeter-client-config
data:
greeter-server-addr: "$(./cluster1 ./scripts/get_service_ip_port.sh default greeter-server)"
EOF
Deploy the greeter client in cluster2:
$ ./cluster2 kubectl apply -k config/cluster2/greeter-client
Create the ClusterSPIFFEID for the greeter server in cluster1:
$ ./cluster1 kubectl apply -f config/greeter-server-id.yaml
Create the ClusterSPIFFEID for the greeter client in cluster2:
$ ./cluster2 kubectl apply -f config/greeter-client-id.yaml
Check the greeter server logs to see that it has received authenticated requests from the greeter client:
$ ./cluster1 kubectl logs deployment/greeter-server
Check the greeter client logs to see that it as able to authenticate the greeter server and issue the request and receive the response:
$ ./cluster2 kubectl logs deployment/greeter-client
List the SPIRE registration entries and federated trust domain relationships that were created by the controller:
$ ./cluster1 scripts/show-spire-entries.sh
$ ./cluster1 scripts/show-spire-federated-bundles.sh
$ ./cluster2 scripts/show-spire-entries.sh
$ ./cluster2 scripts/show-spire-federated-bundles.sh
When you are finished, delete the clusters:
$ ./cluster1 kind delete cluster
$ ./cluster2 kind delete cluster