feat(gobgp): expose BGPProviderService on galactic's gRPC port

cmd/galactic-agent/main.go

@@ -30,19 +30,13 @@ func newRootCommand() *cobra.Command {
 		},
 	}
 
-	cmd.Flags().StringVar(&opts.MetricsAddr, "metrics-addr", ":8082", "Address to serve Prometheus metrics on")
-	cmd.Flags().StringVar(&opts.HealthAddr, "health-addr", ":8083", "Address to serve health/readiness probes on")
 	cmd.Flags().StringVar(&opts.NodeName, "node-name", "", "Override node name (default: NODE_NAME env var)")
-	cmd.Flags().StringVar(&opts.Plane, "plane", "overlay",
-		"BGP plane label published on the BGPProvider (e.g. overlay, overlay-rr)")
-	cmd.Flags().BoolVar(&opts.GoBGPEnabled, "gobgp-enabled", false,
-		"Enable embedded GoBGP and publish BGPProvider")
-	cmd.Flags().IntVar(&opts.GoBGPAPIPort, "gobgp-api-port", 50051,
-		"Port for the embedded GoBGP gRPC API (cosmos dials this)")
-	cmd.Flags().StringVar(&opts.GoBGPLogLevel, "gobgp-log-level", "panic",
-		"GoBGP internal log level (debug, info, warn, error, panic)")
-	cmd.Flags().IntVar(&opts.GRPCHealthPort, "grpc-health-port", 8084,
-		"Port for the gRPC health service (used by Kubernetes readiness probes)")
+	cmd.Flags().StringVar(&opts.Role, "role", "overlay",
+		"Agent role published on the BGPProvider label galactic.io/role (overlay, overlay-rr)")
+	cmd.Flags().IntVar(&opts.Port, "port", 33438,
+		"Port for the gRPC server that cosmos uses to configure the BGP provider")
+	cmd.Flags().IntVar(&opts.HealthPort, "health-port", 5000,
+		"Port for the gRPC health server (Kubernetes liveness and readiness probes)")
 
 	return cmd
 }

deploy/containerlab/resources/bgp/dfw/bgpinstance.yaml

@@ -5,7 +5,7 @@ metadata:
 spec:
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay
+      galactic.io/role: overlay
   asNumber: 65000
   listenPort: 1790
   routerIDSource: Auto

deploy/containerlab/resources/bgp/dfw/bgppeer.yaml

@@ -6,7 +6,7 @@ spec:
   instanceRef: overlay
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay
+      galactic.io/role: overlay
   address: "fc00:0:8::1"
   asNumber: 65000
   remotePort: 1179

deploy/containerlab/resources/bgp/iad/bgpinstance-rr.yaml

@@ -5,7 +5,7 @@ metadata:
 spec:
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay-rr
+      galactic.io/role: overlay-rr
   asNumber: 65000
   routerIDSource: Manual
   routerID: "10.255.255.4"

deploy/containerlab/resources/bgp/iad/bgpinstance.yaml

@@ -5,7 +5,7 @@ metadata:
 spec:
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay
+      galactic.io/role: overlay
   asNumber: 65000
   listenPort: 1790
   routerIDSource: Auto

deploy/containerlab/resources/bgp/iad/bgppeer-rr.yaml

@@ -6,7 +6,7 @@ spec:
   instanceRef: overlay-rr
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay-rr
+      galactic.io/role: overlay-rr
   address: "fc00:0:2::1"
   asNumber: 65000
   passive: true
@@ -23,7 +23,7 @@ spec:
   instanceRef: overlay-rr
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay-rr
+      galactic.io/role: overlay-rr
   address: "fc00:0:3::1"
   asNumber: 65000
   passive: true
@@ -40,7 +40,7 @@ spec:
   instanceRef: overlay-rr
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay-rr
+      galactic.io/role: overlay-rr
   address: "fc00:0:4::1"
   asNumber: 65000
   passive: true

deploy/containerlab/resources/bgp/iad/bgppeer.yaml

@@ -6,7 +6,7 @@ spec:
   instanceRef: overlay
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay
+      galactic.io/role: overlay
   address: "fc00:0:8::1"
   asNumber: 65000
   remotePort: 1179

deploy/containerlab/resources/bgp/sjc/bgpinstance.yaml

@@ -5,7 +5,7 @@ metadata:
 spec:
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay
+      galactic.io/role: overlay
   asNumber: 65000
   listenPort: 1790
   routerIDSource: Auto

deploy/containerlab/resources/bgp/sjc/bgppeer.yaml

@@ -6,7 +6,7 @@ spec:
   instanceRef: overlay
   providerSelector:
     matchLabels:
-      galactic.io/plane: overlay
+      galactic.io/role: overlay
   address: "fc00:0:8::1"
   asNumber: 65000
   remotePort: 1179

deploy/containerlab/resources/cosmos/daemonset-patch.yaml

@@ -0,0 +1,11 @@
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  name: cosmos
+  namespace: bgp-system
+spec:
+  template:
+    spec:
+      containers:
+        - name: cosmos-operator
+          imagePullPolicy: Never

deploy/containerlab/resources/cosmos/kustomization.yaml

@@ -0,0 +1,13 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - ../cosmos-config/deploy/
+images:
+  - name: ghcr.io/milo-os/cosmos
+    newName: cosmos
+    newTag: latest
+patches:
+  - path: daemonset-patch.yaml
+    target:
+      kind: DaemonSet
+      name: cosmos

deploy/containerlab/resources/overlay/base/daemonset.yaml

@@ -26,13 +26,9 @@ spec:
           image: galactic-agent:latest
           imagePullPolicy: Never
           args:
-            - --gobgp-enabled=true
-            - --gobgp-api-port=50051
-            - --gobgp-log-level=debug
-            - --metrics-addr=:8082
-            - --health-addr=:8083
-            - --grpc-health-port=8084
-            - --plane=overlay
+            - --port=33438
+            - --health-port=5000
+            - --role=overlay
           env:
             - name: NODE_NAME
               valueFrom:
@@ -43,14 +39,13 @@ spec:
               add:
                 - NET_ADMIN
           livenessProbe:
-            httpGet:
-              path: /healthz
-              port: 8083
+            grpc:
+              port: 5000
             initialDelaySeconds: 15
             periodSeconds: 20
           readinessProbe:
-            httpGet:
-              path: /readyz
-              port: 8083
+            grpc:
+              port: 5000
+              service: readyz
             initialDelaySeconds: 5
             periodSeconds: 10

deploy/containerlab/resources/overlay/iad/rr/daemonset-patch.yaml

@@ -27,18 +27,13 @@ spec:
       containers:
         - name: galactic-agent
           args:
-            - --gobgp-enabled=true
-            - --gobgp-api-port=50052
-            - --gobgp-log-level=debug
-            - --metrics-addr=:9085
-            - --health-addr=:9086
-            - --grpc-health-port=9087
-            - --plane=overlay-rr
+            - --port=33440
+            - --health-port=5000
+            - --role=overlay-rr
           livenessProbe:
-            httpGet:
-              path: /healthz
-              port: 9086
+            grpc:
+              port: 5000
           readinessProbe:
-            httpGet:
-              path: /readyz
-              port: 9086
+            grpc:
+              port: 5000
+              service: readyz

deploy/containerlab/scripts/install-overlay.sh

@@ -21,12 +21,9 @@ deploy_cosmos() {
   echo "Deploying cosmos operator to ${node}..."
   # Copy the entire config/ directory so that the kustomization's ../crd reference resolves
   docker cp "${COSMOS_DIR}/config" "${node}:/galactic/resources/cosmos-config/"
-  # Use locally built image; patch imagePullPolicy to Never
-  docker exec "${node}" kubectl apply -k /galactic/resources/cosmos-config/deploy/
-  docker exec "${node}" kubectl patch daemonset bgp \
-    -n bgp-system \
-    --type='json' \
-    -p='[{"op":"replace","path":"/spec/template/spec/containers/0/image","value":"cosmos:latest"},{"op":"replace","path":"/spec/template/spec/containers/0/imagePullPolicy","value":"Never"}]'
+  # Copy the local overlay (image + imagePullPolicy overrides for Kind)
+  docker cp "${RESOURCES_DIR}/cosmos" "${node}:/galactic/resources/cosmos-overlay/"
+  docker exec "${node}" kubectl apply -k /galactic/resources/cosmos-overlay/
 }
 
 apply_overlay dfw-control-plane dfw
@@ -37,10 +34,4 @@ deploy_cosmos dfw-control-plane
 deploy_cosmos sjc-control-plane
 deploy_cosmos iad-control-plane
 
-# iad hosts the overlay-rr BGPInstance (route reflector), which requires infra role.
-echo "Patching cosmos clusterRole to infra on iad-control-plane..."
-docker exec iad-control-plane kubectl patch configmap cosmos-config \
-  -n cosmos-system --type=merge -p '{"data":{"clusterRole":"infra"}}'
-docker exec iad-control-plane kubectl rollout restart daemonset bgp -n bgp-system
-
 echo "Done."

deploy/galactic-agent/daemonset.yaml

@@ -22,37 +22,29 @@ spec:
         - name: galactic-agent
           image: ghcr.io/datum-cloud/galactic:latest
           args:
-            - --gobgp-enabled=true
-            - --gobgp-api-port=50051
-            - --gobgp-log-level=panic
-            - --metrics-addr=:8082
-            - --health-addr=:8083
-            - --grpc-health-port=8084
+            - --port=33438
+            - --health-port=5000
           env:
             - name: NODE_NAME
               valueFrom:
                 fieldRef:
                   fieldPath: spec.nodeName
           ports:
-            - name: metrics
-              containerPort: 8082
+            - name: grpc
+              containerPort: 33438
               protocol: TCP
-            - name: healthz
-              containerPort: 8083
-              protocol: TCP
-            - name: grpc-health
-              containerPort: 8084
+            - name: health
+              containerPort: 5000
               protocol: TCP
           livenessProbe:
-            httpGet:
-              path: /healthz
-              port: healthz
+            grpc:
+              port: 5000
             initialDelaySeconds: 10
             periodSeconds: 30
           readinessProbe:
-            httpGet:
-              path: /readyz
-              port: healthz
+            grpc:
+              port: 5000
+              service: readyz
             initialDelaySeconds: 5
             periodSeconds: 10
           securityContext:

docs/agent-startup.md

@@ -6,12 +6,11 @@ sequenceDiagram
     participant GoBGP
     participant Kubernetes
 
-    Agent->>Agent: register Prometheus metrics
-    Agent->>Agent: start gRPC health server (NOT_SERVING)
+    Agent->>Agent: start health gRPC server (--health-port, liveness SERVING immediately)
+    Agent->>Agent: start provider gRPC server (--port, BGPProviderService only)
     Agent->>GoBGP: start embedded server
-    Agent->>GoBGP: WaitReady — poll gRPC API port (30s timeout)
-    Agent->>Kubernetes: EnsureGoBGPProvider (create/update BGPProvider CR)
-    Agent->>Agent: mark gRPC health SERVING
-    Agent->>Agent: mgr.Start — controller-runtime loop
-    Note over Agent: on shutdown: DeleteGoBGPProvider, GracefulStop gRPC
+    Agent->>Kubernetes: EnsureGoBGPProvider (create/update BGPProvider CR with --port address)
+    Agent->>GoBGP: WaitReady — poll in-process API (30s timeout)
+    Agent->>Agent: mark readyz SERVING
+    Note over Agent: on shutdown: mark readyz NOT_SERVING, GracefulStop both gRPC servers
 ```