Stabilize ROS2 integration CI checks

.github/workflows/sanity.yml

@@ -33,5 +33,10 @@ jobs:
         with:
           go-version-file: gimble-ros/go.mod
 
+      - name: Install integration test dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y jq
+
       - name: Run integration tests
         run: ./tests/integration.sh

gimble-ros/internal/ros/graph.go

@@ -8,6 +8,8 @@ import (
 	"time"
 )
 
+var graphCommandTimeout = 5 * time.Second
+
 type TopicConnection struct {
 	Name        string   `json:"name"`
 	Publishers  []string `json:"publishers"`
@@ -33,36 +35,33 @@ func CollectSystemGraph(ctx context.Context, runner Runner) (SystemGraph, error)
 		runner = ExecRunner{}
 	}
 
-	tctx, cancel := WithTimeout(ctx, 5*time.Second)
-	defer cancel()
-
 	graph := SystemGraph{Raw: map[string]string{}}
 
-	nodeList, err := runner.Run(tctx, "ros2", "node", "list")
+	nodeList, err := runGraphCommand(ctx, runner, "ros2", "node", "list")
 	if err != nil {
 		return graph, err
 	}
 	nodes := NonEmptyLines(nodeList)
 	for _, node := range nodes {
-		info, ierr := runner.Run(tctx, "ros2", "node", "info", node)
+		info, ierr := runGraphCommand(ctx, runner, "ros2", "node", "info", node)
 		if ierr != nil {
 			continue
 		}
 		graph.Raw["node_info:"+node] = info
 		graph.Nodes = append(graph.Nodes, ParseNodeDetails(node, info))
 	}
 
-	topicList, _ := runner.Run(tctx, "ros2", "topic", "list")
+	topicList, _ := runGraphCommand(ctx, runner, "ros2", "topic", "list")
 	for _, topic := range NonEmptyLines(topicList) {
-		info, ierr := runner.Run(tctx, "ros2", "topic", "info", "-v", topic)
+		info, ierr := runGraphCommand(ctx, runner, "ros2", "topic", "info", "-v", topic)
 		if ierr != nil {
 			continue
 		}
 		graph.Raw["topic_info:"+topic] = info
 		graph.Topics = append(graph.Topics, ParseTopicInfo(topic, info))
 	}
 
-	svcList, _ := runner.Run(tctx, "ros2", "service", "list")
+	svcList, _ := runGraphCommand(ctx, runner, "ros2", "service", "list")
 	graph.Services = NonEmptyLines(svcList)
 	sort.Strings(graph.Services)
 
@@ -152,3 +151,9 @@ func NonEmptyLines(in string) []string {
 	}
 	return out
 }
+
+func runGraphCommand(ctx context.Context, runner Runner, name string, args ...string) (string, error) {
+	tctx, cancel := WithTimeout(ctx, graphCommandTimeout)
+	defer cancel()
+	return runner.Run(tctx, name, args...)
+}

gimble-ros/internal/ros/graph_timeout_test.go

@@ -0,0 +1,74 @@
+package ros
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"testing"
+	"time"
+)
+
+type delayedGraphRunner struct {
+	delay  time.Duration
+	output map[string]string
+}
+
+func (r delayedGraphRunner) Run(ctx context.Context, name string, args ...string) (string, error) {
+	select {
+	case <-time.After(r.delay):
+	case <-ctx.Done():
+		return "", ctx.Err()
+	}
+	key := strings.Join(append([]string{name}, args...), " ")
+	if out, ok := r.output[key]; ok {
+		return out, nil
+	}
+	return "", fmt.Errorf("unexpected command: %s", key)
+}
+
+func TestCollectSystemGraph_UsesPerCommandTimeout(t *testing.T) {
+	old := graphCommandTimeout
+	graphCommandTimeout = 20 * time.Millisecond
+	t.Cleanup(func() { graphCommandTimeout = old })
+
+	runner := delayedGraphRunner{
+		delay: 12 * time.Millisecond,
+		output: map[string]string{
+			"ros2 node list": "/talker\n/listener",
+			"ros2 node info /talker": `Publishers:
+  * /chatter [std_msgs/msg/String]
+Subscribers:
+  * /parameter_events [rcl_interfaces/msg/ParameterEvent]
+Service Servers:
+  * /talker/get_parameters`,
+			"ros2 node info /listener": `Publishers:
+  * /parameter_events [rcl_interfaces/msg/ParameterEvent]
+Subscribers:
+  * /chatter [std_msgs/msg/String]
+Service Servers:
+  * /listener/get_parameters`,
+			"ros2 topic list": "/chatter",
+			"ros2 topic info -v /chatter": `Publisher count:
+Publishers:
+  Node name: /talker
+Subscription count:
+Subscriptions:
+  Node name: /listener`,
+			"ros2 service list": "/talker/get_parameters\n/listener/get_parameters",
+		},
+	}
+
+	graph, err := CollectSystemGraph(context.Background(), runner)
+	if err != nil {
+		t.Fatalf("CollectSystemGraph returned error: %v", err)
+	}
+	if len(graph.Nodes) != 2 {
+		t.Fatalf("expected 2 nodes, got %d", len(graph.Nodes))
+	}
+	if len(graph.Topics) != 1 {
+		t.Fatalf("expected 1 topic, got %d", len(graph.Topics))
+	}
+	if len(graph.Services) != 2 {
+		t.Fatalf("expected 2 services, got %d", len(graph.Services))
+	}
+}

tests/integration.sh

@@ -13,16 +13,37 @@ cleanup() {
   docker stop "$CONTAINER_NAME" 2>/dev/null || true
   docker rm "$CONTAINER_NAME" 2>/dev/null || true
 }
+
+print_debug_info() {
+  echo "==> Debug: container status"
+  docker ps -a --filter "name=$CONTAINER_NAME" || true
+  echo "==> Debug: container logs"
+  docker logs "$CONTAINER_NAME" 2>/dev/null || true
+}
+on_error() {
+  echo "ERROR: integration script failed at line $1"
+  print_debug_info
+}
+trap 'on_error $LINENO' ERR
 trap cleanup EXIT
 
 cd "$REPO_ROOT"
 
+echo "==> Ensuring no stale container with same name exists..."
+docker rm -f "$CONTAINER_NAME" 2>/dev/null || true
+
 echo "==> Building ROS2 demo image..."
 docker build -t "$IMAGE_NAME" -f gimble-ros/docker/Dockerfile .
 
 echo "==> Starting ROS2 container..."
 docker run -d --name "$CONTAINER_NAME" "$IMAGE_NAME"
 
+if [[ "$(docker inspect -f '{{.State.Running}}' "$CONTAINER_NAME" 2>/dev/null || echo false)" != "true" ]]; then
+  echo "ERROR: ROS2 container is not running"
+  print_debug_info
+  exit 1
+fi
+
 echo "==> Waiting for ROS2 nodes to start..."
 for i in {1..30}; do
   if docker exec "$CONTAINER_NAME" ros2 node list 2>/dev/null | grep -q "/talker"; then
@@ -32,6 +53,7 @@ for i in {1..30}; do
   if [[ $i -eq 30 ]]; then
     echo "ERROR: /talker node did not appear within 30s"
     docker exec "$CONTAINER_NAME" ros2 node list 2>/dev/null || true
+    print_debug_info
     exit 1
   fi
   sleep 1
@@ -43,6 +65,12 @@ make build
 echo "==> Running gim graph ros2 --container $CONTAINER_NAME..."
 GRAPH_JSON=$(./gimble-ros/gim graph ros2 --container "$CONTAINER_NAME")
 
+if ! echo "$GRAPH_JSON" | jq -e . >/dev/null; then
+  echo "ERROR: gim graph did not return valid JSON"
+  echo "$GRAPH_JSON"
+  exit 1
+fi
+
 echo "==> Verifying graph output..."
 NODE_COUNT=$(echo "$GRAPH_JSON" | jq -r '.nodes | length')
 TOPIC_COUNT=$(echo "$GRAPH_JSON" | jq -r '.topics | length')
@@ -62,7 +90,7 @@ fi
 echo "    Found $NODE_COUNT node(s), $TOPIC_COUNT topic(s)"
 
 echo "==> Running gim analyze ros2 --container $CONTAINER_NAME..."
-ANALYZE_OUT=$(./gimble-ros/gim analyze ros2 --container "$CONTAINER_NAME" 2>&1)
+ANALYZE_OUT=$(GIMBLE_AI_API_KEY= ./gimble-ros/gim analyze ros2 --container "$CONTAINER_NAME" 2>&1)
 
 if [[ -z "$ANALYZE_OUT" ]]; then
   echo "ERROR: gim analyze produced no output"