Collect Gloo metrics and some snapshots on test failure (#10400)

Co-authored-by: changelog-bot <changelog-bot>
Co-authored-by: Nathan Fudenberg <[email protected]>

Author: ryanrolds

changelog/v1.18.0-rc3/collect-more-artifacts-on-ci-failure.yaml

@@ -0,0 +1,14 @@
+changelog:
+  - type: NON_USER_FACING
+    description: >-
+      Gloo Gateway controller metrics and xds/krt snaphots are now collected and included 
+      the test failure artifacts. 
+      After encountering some test failures that proved difficult to debug without knowing more
+      about the state of the cluster, we have added additional artifacts to be collected when
+      a test fails.
+      This will help us to more easily diagnose the cause of test failures.
+  - type: NON_USER_FACING
+    description: >-
+      Unified the kube2e and kubernetes/e2e test failure artifact collection.
+      Previously, the test failure artifacts for kube2e and kubernetes/e2e tests were different
+      and produced by their own logic. 

pkg/utils/glooadminutils/admincli/client.go

@@ -14,6 +14,8 @@ import (
 
 const (
 	InputSnapshotPath = "/snapshots/input"
+	xdsSnapshotPath   = "/snapshots/xds"
+	krtSnapshotPath   = "/snapshots/krt"
 )
 
 // Client is a utility for executing requests against the Gloo Admin API
@@ -84,6 +86,16 @@ func (c *Client) InputSnapshotCmd(ctx context.Context) cmdutils.Cmd {
 	return c.Command(ctx, curl.WithPath(InputSnapshotPath))
 }
 
+// XdsSnapshotCmd returns the cmdutils.Cmd that can be run, and will execute a request against the XDS Snapshot path
+func (c *Client) XdsSnapshotCmd(ctx context.Context) cmdutils.Cmd {
+	return c.Command(ctx, curl.WithPath(xdsSnapshotPath))
+}
+
+// KrtSnapshotCmd returns the cmdutils.Cmd that can be run, and will execute a request against the KRT Snapshot path
+func (c *Client) KrtSnapshotCmd(ctx context.Context) cmdutils.Cmd {
+	return c.Command(ctx, curl.WithPath(krtSnapshotPath))
+}
+
 // GetInputSnapshot returns the data that is available at the input snapshot endpoint
 func (c *Client) GetInputSnapshot(ctx context.Context) ([]interface{}, error) {
 	var outLocation threadsafe.Buffer

pkg/utils/kubeutils/kubectl/cli.go

@@ -342,3 +342,28 @@ func (c *Cli) GetContainerLogs(ctx context.Context, namespace string, name strin
 	stdout, stderr, err := c.Execute(ctx, "-n", namespace, "logs", name)
 	return stdout + stderr, err
 }
+
+// GetPodsInNsWithLabel returns the pods in the specified namespace with the specified label
+func (c *Cli) GetPodsInNsWithLabel(ctx context.Context, namespace string, label string) ([]string, error) {
+	podStdOut := bytes.NewBuffer(nil)
+	podStdErr := bytes.NewBuffer(nil)
+
+	// Fetch the name of the Gloo Gateway controller pod
+	getGlooPodNamesCmd := c.Command(ctx, "get", "pod", "-n", namespace,
+		"--selector", label, "--output", "jsonpath='{.items[*].metadata.name}'")
+	err := getGlooPodNamesCmd.WithStdout(podStdOut).WithStderr(podStdErr).Run().Cause()
+	if err != nil {
+		fmt.Printf("error running get gloo pod name command: %v\n", err)
+	}
+
+	// Clean up and check the output
+	glooPodNamesString := strings.Trim(podStdOut.String(), "'")
+	if glooPodNamesString == "" {
+		fmt.Printf("no %s pods found in namespace %s\n", label, namespace)
+		return []string{}, nil
+	}
+
+	// Split the string on whitespace to get the pod names
+	glooPodNames := strings.Fields(glooPodNamesString)
+	return glooPodNames, nil
+}

projects/gloo/pkg/servers/iosnapshot/history_test.go

@@ -1,4 +1,4 @@
-package iosnapshot
+package iosnapshot_test
 
 import (
 	"context"
@@ -8,6 +8,7 @@ import (
 	gomegatypes "github.com/onsi/gomega/types"
 	"github.com/solo-io/gloo/pkg/schemes"
 	gloov1 "github.com/solo-io/gloo/projects/gloo/pkg/api/v1/kube/apis/gloo.solo.io/v1"
+	"github.com/solo-io/gloo/projects/gloo/pkg/servers/iosnapshot"
 	apiv1beta1 "sigs.k8s.io/gateway-api/apis/v1beta1"
 
 	wellknownkube "github.com/solo-io/gloo/projects/gloo/pkg/api/v1/kube/wellknown"
@@ -52,16 +53,16 @@ var _ = Describe("History", func() {
 		ctx context.Context
 
 		clientBuilder *fake.ClientBuilder
-		history       History
+		history       iosnapshot.History
 
-		historyFactorParams HistoryFactoryParameters
+		historyFactorParams iosnapshot.HistoryFactoryParameters
 	)
 
 	BeforeEach(func() {
 		ctx = context.Background()
 		clientBuilder = fake.NewClientBuilder().WithScheme(schemes.DefaultScheme())
 
-		historyFactorParams = HistoryFactoryParameters{
+		historyFactorParams = iosnapshot.HistoryFactoryParameters{
 			Settings: &v1.Settings{
 				Metadata: &core.Metadata{
 					Name:      "my-settings",
@@ -98,11 +99,11 @@ var _ = Describe("History", func() {
 					},
 				}
 
-				history = NewHistory(
+				history = iosnapshot.NewHistory(
 					historyFactorParams.Cache,
 					historyFactorParams.Settings,
 					clientBuilder.WithObjects(clientObjects...).Build(),
-					append(CompleteInputSnapshotGVKs, deploymentGvk), // include the Deployment GVK
+					append(iosnapshot.CompleteInputSnapshotGVKs, deploymentGvk), // include the Deployment GVK
 				)
 			})
 
@@ -136,15 +137,15 @@ var _ = Describe("History", func() {
 					},
 				}
 
-				history = NewHistory(&xds.MockXdsCache{},
+				history = iosnapshot.NewHistory(&xds.MockXdsCache{},
 					&v1.Settings{
 						Metadata: &core.Metadata{
 							Name:      "my-settings",
 							Namespace: defaults.GlooSystem,
 						},
 					},
 					clientBuilder.WithObjects(clientObjects...).Build(),
-					CompleteInputSnapshotGVKs, // do not include the Deployment GVK
+					iosnapshot.CompleteInputSnapshotGVKs, // do not include the Deployment GVK
 				)
 			})
 
@@ -377,11 +378,11 @@ var _ = Describe("History", func() {
 				},
 			}
 
-			history = NewHistory(
+			history = iosnapshot.NewHistory(
 				historyFactorParams.Cache,
 				historyFactorParams.Settings,
 				clientBuilder.WithObjects(clientObjects...).Build(),
-				CompleteInputSnapshotGVKs)
+				iosnapshot.CompleteInputSnapshotGVKs)
 		})
 
 		Context("Kubernetes Core Resources", func() {
@@ -663,11 +664,11 @@ var _ = Describe("History", func() {
 	Context("GetEdgeApiSnapshot", func() {
 
 		BeforeEach(func() {
-			history = NewHistory(
+			history = iosnapshot.NewHistory(
 				historyFactorParams.Cache,
 				historyFactorParams.Settings,
 				clientBuilder.Build(), // no objects, because this API doesn't rely on the kube client
-				CompleteInputSnapshotGVKs,
+				iosnapshot.CompleteInputSnapshotGVKs,
 			)
 		})
 
@@ -758,11 +759,11 @@ var _ = Describe("History", func() {
 	Context("GetProxySnapshot", func() {
 
 		BeforeEach(func() {
-			history = NewHistory(
+			history = iosnapshot.NewHistory(
 				historyFactorParams.Cache,
 				historyFactorParams.Settings,
 				clientBuilder.Build(), // no objects, because this API doesn't rely on the kube client
-				CompleteInputSnapshotGVKs,
+				iosnapshot.CompleteInputSnapshotGVKs,
 			)
 		})
 
@@ -794,7 +795,7 @@ var _ = Describe("History", func() {
 
 })
 
-func getInputSnapshotObjects(ctx context.Context, history History) []client.Object {
+func getInputSnapshotObjects(ctx context.Context, history iosnapshot.History) []client.Object {
 	snapshotResponse := history.GetInputSnapshot(ctx)
 	Expect(snapshotResponse.Error).NotTo(HaveOccurred())
 
@@ -804,7 +805,7 @@ func getInputSnapshotObjects(ctx context.Context, history History) []client.Obje
 	return responseObjects
 }
 
-func getProxySnapshotResources(ctx context.Context, history History) []crdv1.Resource {
+func getProxySnapshotResources(ctx context.Context, history iosnapshot.History) []crdv1.Resource {
 	snapshotResponse := history.GetProxySnapshot(ctx)
 	Expect(snapshotResponse.Error).NotTo(HaveOccurred())
 
@@ -814,7 +815,7 @@ func getProxySnapshotResources(ctx context.Context, history History) []crdv1.Res
 	return responseObjects
 }
 
-func getEdgeApiSnapshot(ctx context.Context, history History) *v1snap.ApiSnapshot {
+func getEdgeApiSnapshot(ctx context.Context, history iosnapshot.History) *v1snap.ApiSnapshot {
 	snapshotResponse := history.GetEdgeApiSnapshot(ctx)
 	Expect(snapshotResponse.Error).NotTo(HaveOccurred())
 
@@ -827,7 +828,7 @@ func getEdgeApiSnapshot(ctx context.Context, history History) *v1snap.ApiSnapsho
 // setSnapshotOnHistory sets the ApiSnapshot on the history, and blocks until it has been processed
 // This is a utility method to help developers write tests, without having to worry about the asynchronous
 // nature of the `Set` API on the History
-func setSnapshotOnHistory(ctx context.Context, history History, snap *v1snap.ApiSnapshot) {
+func setSnapshotOnHistory(ctx context.Context, history iosnapshot.History, snap *v1snap.ApiSnapshot) {
 	gwSignal := &gatewayv1.Gateway{
 		// We append a custom Gateway to the Snapshot, and then use that object
 		// to verify the Snapshot has been processed