test/framework: add functions to collect infrastructure logs in tests, e.g. from CAPD's lb container

Author: chrischdi

docs/book/src/developer/providers/migrations/v1.4-to-v1.5.md

@@ -31,6 +31,7 @@ maintainers of providers and consumers of our Go API.
 
 - clusterctl move is adding the new annotation `clusterctl.cluster.x-k8s.io/delete-for-move` before object deletion.
 - Providers running CAPI release-0.3 clusterctl upgrade tests should set `WorkloadKubernetesVersion` field to the maximum workload cluster kubernetes version supported by the old providers in `ClusterctlUpgradeSpecInput`. For more information, please see: https://github.com/kubernetes-sigs/cluster-api/pull/8518#issuecomment-1508064859 
+- Introduced function `CollectInfrastructureLogs` at the `ClusterLogCollector` interface in `test/framework/cluster_proxy.go` to allow collecting infrastructure related logs during tests.
 
 ### Suggested changes for providers
 

test/framework/cluster_proxy.go

@@ -93,7 +93,7 @@ type ClusterProxy interface {
 	// GetWorkloadCluster returns a proxy to a workload cluster defined in the Kubernetes cluster.
 	GetWorkloadCluster(ctx context.Context, namespace, name string) ClusterProxy
 
-	// CollectWorkloadClusterLogs collects machines logs from the workload cluster.
+	// CollectWorkloadClusterLogs collects machines and infrastructure logs from the workload cluster.
 	CollectWorkloadClusterLogs(ctx context.Context, namespace, name, outputPath string)
 
 	// Dispose proxy's internal resources (the operation does not affects the Kubernetes cluster).
@@ -107,6 +107,8 @@ type ClusterLogCollector interface {
 	// TODO: describe output folder struct
 	CollectMachineLog(ctx context.Context, managementClusterClient client.Client, m *clusterv1.Machine, outputPath string) error
 	CollectMachinePoolLog(ctx context.Context, managementClusterClient client.Client, m *expv1.MachinePool, outputPath string) error
+	// CollectInfrastructureLogs collects log from the infrastructure.
+	CollectInfrastructureLogs(ctx context.Context, managementClusterClient client.Client, c *clusterv1.Cluster, outputPath string) error
 }
 
 // Option is a configuration option supplied to NewClusterProxy.
@@ -280,7 +282,7 @@ func (p *clusterProxy) GetWorkloadCluster(ctx context.Context, namespace, name s
 	return newFromAPIConfig(name, config, p.scheme)
 }
 
-// CollectWorkloadClusterLogs collects machines logs from the workload cluster.
+// CollectWorkloadClusterLogs collects machines and infrastructure logs and from the workload cluster.
 func (p *clusterProxy) CollectWorkloadClusterLogs(ctx context.Context, namespace, name, outputPath string) {
 	if p.logCollector == nil {
 		return
@@ -317,6 +319,23 @@ func (p *clusterProxy) CollectWorkloadClusterLogs(ctx context.Context, namespace
 			fmt.Printf("Failed to get logs for MachinePool %s, Cluster %s: %v\n", mp.GetName(), klog.KRef(namespace, name), err)
 		}
 	}
+
+	cluster := &clusterv1.Cluster{}
+	Eventually(func() error {
+		key := client.ObjectKey{
+			Namespace: namespace,
+			Name:      name,
+		}
+		return client.IgnoreNotFound(p.GetClient().Get(ctx, key, cluster))
+	}, retryableOperationTimeout, retryableOperationInterval).Should(Succeed(), "Failed to get Cluster %s", klog.KRef(namespace, name))
+
+	if cluster != nil {
+		err := p.logCollector.CollectInfrastructureLogs(ctx, p.GetClient(), cluster, path.Join(outputPath, "infrastructure"))
+		if err != nil {
+			// NB. we are treating failures in collecting logs as a non blocking operation (best effort)
+			fmt.Printf("Failed to get infrastructure logs for Cluster %s: %v\n", klog.KRef(namespace, name), err)
+		}
+	}
 }
 
 func getMachinesInCluster(ctx context.Context, c client.Client, namespace, name string) (*clusterv1.MachineList, error) {

test/framework/docker_logcollector.go

@@ -76,6 +76,25 @@ func (k DockerLogCollector) CollectMachinePoolLog(ctx context.Context, _ client.
 	return kerrors.NewAggregate(errs)
 }
 
+func (k DockerLogCollector) CollectInfrastructureLogs(ctx context.Context, _ client.Client, c *clusterv1.Cluster, outputPath string) error {
+	containerRuntime, err := container.NewDockerClient()
+	if err != nil {
+		return err
+	}
+	ctx = container.RuntimeInto(ctx, containerRuntime)
+
+	lbContainerName := fmt.Sprintf("%s-lb", c.GetName())
+
+	f, err := fileOnHost(filepath.Join(outputPath, fmt.Sprintf("%s.log", lbContainerName)))
+	if err != nil {
+		return err
+	}
+
+	defer f.Close()
+
+	return containerRuntime.ContainerDebugInfo(ctx, lbContainerName, f)
+}
+
 func (k DockerLogCollector) collectLogsFromNode(ctx context.Context, outputPath string, containerName string) error {
 	containerRuntime, err := container.RuntimeFrom(ctx)
 	if err != nil {

test/infrastructure/container/docker.go

@@ -22,6 +22,7 @@ import (
 	"bytes"
 	"context"
 	"encoding/csv"
+	"encoding/json"
 	"fmt"
 	"io"
 	"os"
@@ -331,8 +332,13 @@ func (d *dockerRuntime) ContainerDebugInfo(ctx context.Context, containerName st
 		return errors.Wrapf(err, "failed to inspect container %q", containerName)
 	}
 
+	rawJSON, err := json.Marshal(containerInfo)
+	if err != nil {
+		return errors.Wrapf(err, "failed to marshal container info to json")
+	}
+
 	fmt.Fprintln(w, "Inspected the container:")
-	fmt.Fprintf(w, "%+v\n", containerInfo)
+	fmt.Fprintf(w, "%s\n", rawJSON)
 
 	options := types.ContainerLogsOptions{
 		ShowStdout: true,