Added e2e test to reproduce issue #340 (#397)

* Added e2e tests to emulate the condition.
Minor fixes in the e2e script.

* Incorporated feedback and tests from #398

* Added unit tests to emulate a long running consumer

Author: z103cb

go.mod

@@ -3,6 +3,7 @@ module github.com/project-codeflare/multi-cluster-app-dispatcher
 go 1.18
 
 require (
+	github.com/eapache/go-resiliency v1.3.0
 	github.com/emicklei/go-restful v2.16.0+incompatible
 	github.com/golang/protobuf v1.4.3
 	github.com/hashicorp/go-multierror v1.1.1

go.sum

@@ -96,6 +96,8 @@ github.com/docopt/docopt-go v0.0.0-20180111231733-ee0de3bc6815/go.mod h1:WwZ+bS3
 github.com/dustin/go-humanize v0.0.0-20171111073723-bb3d318650d4/go.mod h1:HtrtbFcZ19U5GC7JDqmcUSB87Iq5E25KnS6fMYU6eOk=
 github.com/dustin/go-humanize v1.0.0 h1:VSnTsYCnlFHaM2/igO1h6X3HA71jcobQuxemgkq4zYo=
 github.com/dustin/go-humanize v1.0.0/go.mod h1:HtrtbFcZ19U5GC7JDqmcUSB87Iq5E25KnS6fMYU6eOk=
+github.com/eapache/go-resiliency v1.3.0 h1:RRL0nge+cWGlxXbUzJ7yMcq6w2XBEr19dCN6HECGaT0=
+github.com/eapache/go-resiliency v1.3.0/go.mod h1:5yPzW0MIvSe0JDsv0v+DvcjEv2FyD6iZYSs1ZI+iQho=
 github.com/elazarl/goproxy v0.0.0-20180725130230-947c36da3153/go.mod h1:/Zj4wYkgs4iZTTu3o/KG3Itv/qCCa8VVMlb3i9OVuzc=
 github.com/emicklei/go-restful v0.0.0-20170410110728-ff4f55a20633/go.mod h1:otzb+WCGbkyDHkqmQmT5YD2WR4BBwUdeQoFo8l/7tVs=
 github.com/emicklei/go-restful v2.9.5+incompatible/go.mod h1:otzb+WCGbkyDHkqmQmT5YD2WR4BBwUdeQoFo8l/7tVs=

hack/run-e2e-kind.sh

@@ -33,6 +33,7 @@ export CLUSTER_CONTEXT="--name test"
 # Using older image due to older version of kubernetes cluster"
 export IMAGE_ECHOSERVER="kicbase/echo-server:1.0"
 export IMAGE_UBUNTU_LATEST="ubuntu:latest"
+export IMAGE_UBI_LATEST="registry.access.redhat.com/ubi8/ubi:latest"
 export KIND_OPT=${KIND_OPT:=" --config ${ROOT_DIR}/hack/e2e-kind-config.yaml"}
 export KA_BIN=_output/bin
 export WAIT_TIME="20s"
@@ -220,6 +221,13 @@ function kind-up-cluster {
     exit 1
   fi
 
+  docker pull ${IMAGE_UBI_LATEST}
+  if [ $? -ne 0 ]
+  then
+    echo "Failed to pull ${IMAGE_UBI_LATEST}"
+    exit 1
+  fi
+
   if [[ "$MCAD_IMAGE_PULL_POLICY" = "Always" ]]
   then
     docker pull ${IMAGE_MCAD}
@@ -236,7 +244,7 @@ function kind-up-cluster {
   fi
   docker images
 
-  for image in ${IMAGE_ECHOSERVER} ${IMAGE_UBUNTU_LATEST} ${IMAGE_MCAD}
+  for image in ${IMAGE_ECHOSERVER} ${IMAGE_UBUNTU_LATEST} ${IMAGE_MCAD} ${IMAGE_UBI_LATEST}
   do
     kind load docker-image ${image} ${CLUSTER_CONTEXT}
     if [ $? -ne 0 ]

pkg/controller/quota/quotaforestmanager/qm_lib_backend_with_quotasubt_mgr.go

@@ -530,20 +530,19 @@ func (qm *QuotaManager) Fits(aw *arbv1.AppWrapper, awResDemands *clusterstateapi
 	klog.V(4).Infof("[Fits] Sending quota allocation request: %#v ", consumerInfo)
 	allocResponse, err := qm.quotaManagerBackend.AllocateForest(QuotaManagerForestName, consumerID)
 	if err != nil {
+		qm.removeConsumer(consumerID)
 		klog.Errorf("[Fits] Error allocating consumer: %s/%s, err=%#v.", aw.Namespace, aw.Name, err)
 		return false, nil, err.Error()
 	}
-	if allocResponse != nil && len(strings.TrimSpace(allocResponse.GetMessage())) > 0 {
-		klog.Errorf("[Fits] Error allocating consumer: %s/%s, msg=%s, err=%#v.",
-			aw.Namespace, aw.Name, allocResponse.GetMessage(), err)
-		return false, nil, allocResponse.GetMessage()
-	}
 	klog.V(4).Infof("[Fits] allocation response received. Quota Allocated: %t, Message: '%s', Preempted app wrappers count: %d", allocResponse.IsAllocated(),
 		strings.TrimSpace(allocResponse.GetMessage()), len(allocResponse.GetPreemptedIds()))
 	doesFit := allocResponse.IsAllocated()
+	if !doesFit {
+		qm.removeConsumer(consumerID)
+		return doesFit, preemptIds, strings.TrimSpace(allocResponse.GetMessage())
+	}
 	preemptIds = qm.getAppWrappers(allocResponse.GetPreemptedIds())
-
-	return doesFit, preemptIds, ""
+	return doesFit, preemptIds, strings.TrimSpace(allocResponse.GetMessage())
 }
 
 func (qm *QuotaManager) getAppWrappers(preemptIds []string) []*arbv1.AppWrapper {
@@ -614,3 +613,14 @@ func (qm *QuotaManager) Release(aw *arbv1.AppWrapper) bool {
 
 	return released
 }
+func (qm *QuotaManager) removeConsumer(consumerID string) {
+	// removing the consumer to allow for the consumer to be added if and when
+	// the function is called for the same app wrapper
+	removed, err := qm.quotaManagerBackend.RemoveConsumer(consumerID)
+	if err != nil {
+		klog.Warningf("Failed to remove consumer %s, %#v", consumerID, err)
+	}
+	if !removed {
+		klog.Warningf("Failed to remove consumer %s", consumerID)
+	}
+}

pkg/quotaplugins/quota-forest/quota-manager/quota/quotamanager_test.go

@@ -19,7 +19,9 @@ package quota_test
 import (
 	"strings"
 	"testing"
+	"time"
 
+	"github.com/eapache/go-resiliency/retrier"
 	"github.com/project-codeflare/multi-cluster-app-dispatcher/pkg/quotaplugins/quota-forest/quota-manager/quota"
 	"github.com/project-codeflare/multi-cluster-app-dispatcher/pkg/quotaplugins/quota-forest/quota-manager/quota/utils"
 	"github.com/stretchr/testify/assert"
@@ -225,3 +227,169 @@ func TestNewQuotaManagerConsumerAllocationRelease(t *testing.T) {
 		})
 	}
 }
+func TestQuotaManagerQuotaUsedLongRunningConsumers(t *testing.T) {
+	forestName := "unit-test-1"
+	qmManagerUnderTest := quota.NewManager()
+
+	err := qmManagerUnderTest.AddForest(forestName)
+	assert.NoError(t, err, "No error expected when adding a forest")
+	testTreeName, err := qmManagerUnderTest.AddTreeFromString(
+		`{
+			"kind": "QuotaTree",
+			"metadata": {
+			  "name": "test-tree"
+			},
+			"spec": {
+			  "resourceNames": [
+				"cpu",
+				"memory"
+			  ],
+			  "nodes": {		
+				"root": {
+				  "parent": "nil",
+				  "hard": "true",
+				  "quota": {
+					"cpu": "10",
+					"memory": "256"
+				  }
+				},		
+				"gold": {
+				  "parent": "root",
+				  "hard": "true",
+				  "quota": {
+					"cpu": "10",
+					"memory": "256"
+				  }
+				}
+			  }
+			}
+		}`)
+	if err != nil {
+		assert.Fail(t, "No error expected when adding a tree to forest")
+	}
+	err = qmManagerUnderTest.AddTreeToForest(forestName, testTreeName)
+	assert.NoError(t, err, "No error expected when adding a tree from forest")
+	modeSet := qmManagerUnderTest.SetMode(quota.Normal)
+	assert.True(t, modeSet, "Setting the mode should not fail.")
+
+	// Define the test table
+	var tests = []struct {
+		name     string
+		consumer utils.JConsumer
+	}{
+		// the table itself
+		{"Gold consumer 1",
+			utils.JConsumer{
+				Kind: "Consumer",
+				MetaData: utils.JMetaData{
+					Name: "gpld-consumer-data",
+				},
+				Spec: utils.JConsumerSpec{
+					ID: "gold-consumer-1",
+					Trees: []utils.JConsumerTreeSpec{
+						{
+							TreeName: testTreeName,
+							GroupID:  "gold",
+							Request: map[string]int{
+								"cpu":    10,
+								"memory": 4,
+								"gpu":    0,
+							},
+							Priority:      0,
+							CType:         0,
+							UnPreemptable: false,
+						},
+					},
+				},
+			},
+		},
+		// the table itself
+		{"Gold consumer 2",
+			utils.JConsumer{
+				Kind: "Consumer",
+				MetaData: utils.JMetaData{
+					Name: "gpld-consumer-data",
+				},
+				Spec: utils.JConsumerSpec{
+					ID: "gold-consumer-2",
+					Trees: []utils.JConsumerTreeSpec{
+						{
+							TreeName: testTreeName,
+							GroupID:  "gold",
+							Request: map[string]int{
+								"cpu":    10,
+								"memory": 4,
+								"gpu":    0,
+							},
+							Priority:      0,
+							CType:         0,
+							UnPreemptable: false,
+						},
+					},
+				},
+			},
+		},
+	}
+	// Execute tests in parallel
+	for _, tc := range tests {
+		tc := tc // capture range variable
+		t.Run(tc.name, func(t *testing.T) {
+			t.Parallel()
+			// Get list of quota management tree IDs
+			qmTreeIDs := qmManagerUnderTest.GetTreeNames()
+
+			consumerInfo, err := quota.NewConsumerInfo(tc.consumer)
+			assert.NoError(t, err, "No error expected when building consumer")
+			assert.Contains(t, qmTreeIDs, tc.consumer.Spec.Trees[0].TreeName)
+
+			retryAllocation := retrier.New(retrier.LimitedExponentialBackoff(10, 10*time.Millisecond, 1*time.Second),
+				&AllocationClassifier{})
+
+			err = retryAllocation.Run(func() error {
+				added, err2 := qmManagerUnderTest.AddConsumer(consumerInfo)
+				assert.NoError(t, err2, "No error expected when adding consumer")
+				assert.True(t, added, "Consumer is expected to be added")
+
+				response, err2 := qmManagerUnderTest.AllocateForest(forestName, consumerInfo.GetID())
+				if err2 == nil {
+					assert.Equal(t, 0, len(strings.TrimSpace(response.GetMessage())), "A empty response is expected")
+					assert.True(t, response.IsAllocated(), "The allocation should succeed")
+				} else {
+					removed, err3 := qmManagerUnderTest.RemoveConsumer(consumerInfo.GetID())
+					assert.NoError(t, err3, "No Error expected when removing consumer")
+					assert.True(t, removed, "Removal of consumer should succeed")
+				}
+				return err2
+
+			})
+			if err != nil {
+				assert.Failf(t, "Allocation of quota should have succeed: '%s'", err.Error())
+			}
+
+			//simulate a long running consumer that has quota allocated
+			time.Sleep(10 * time.Millisecond)
+
+			deAllocated := qmManagerUnderTest.DeAllocateForest(forestName, consumerInfo.GetID())
+			assert.True(t, deAllocated, "De-allocation expected to succeed")
+
+			removed, err := qmManagerUnderTest.RemoveConsumer(consumerInfo.GetID())
+			assert.NoError(t, err, "No Error expected when removing consumer")
+			assert.True(t, removed, "Removal of consumer should succeed")
+
+		})
+	}
+
+}
+
+type AllocationClassifier struct {
+}
+
+func (c *AllocationClassifier) Classify(err error) retrier.Action {
+	if err == nil {
+		return retrier.Succeed
+	}
+	if strings.TrimSpace(err.Error()) == "Failed to allocate quota on quota designation 'test-tree'" {
+		return retrier.Retry
+	}
+	return retrier.Fail
+}

test/e2e-kuttl/quota-errors/00-assert.yaml

@@ -0,0 +1,25 @@
+# Verify CRDs existence
+apiVersion: apiextensions.k8s.io/v1
+kind: CustomResourceDefinition
+metadata:
+  name: appwrappers.mcad.ibm.com
+status:
+  acceptedNames:
+    kind: AppWrapper
+    listKind: AppWrapperList
+    plural: appwrappers
+    singular: appwrapper
+  storedVersions:
+    - v1beta1
+---
+apiVersion: apiextensions.k8s.io/v1
+kind: CustomResourceDefinition
+metadata:
+  name: quotasubtrees.ibm.com
+status:
+  acceptedNames:
+    kind: QuotaSubtree
+    singular: quotasubtree
+    plural: quotasubtrees
+  storedVersions:
+    - v1

test/e2e-kuttl/quota-errors/01-assert.yaml

@@ -0,0 +1,24 @@
+# Verify subtree creations
+apiVersion: ibm.com/v1
+kind: QuotaSubtree
+metadata:
+  name: context-root
+  namespace: kube-system
+  labels:
+    tree: quota_context
+---
+apiVersion: ibm.com/v1
+kind: QuotaSubtree
+metadata:
+  name: service-root
+  namespace: kube-system
+  labels:
+    tree: quota_service
+---
+apiVersion: ibm.com/v1
+kind: QuotaSubtree
+metadata:
+  name: context-root-children
+  namespace: kube-system
+  labels:
+    tree: quota_context

test/e2e-kuttl/quota-errors/02-assert.yaml

@@ -0,0 +1,5 @@
+# Verify test namespace existence
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: quota-errors

test/e2e-kuttl/quota-errors/02-install.yaml

@@ -0,0 +1,4 @@
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: quota-errors

test/e2e-kuttl/quota-errors/03-assert.yaml

@@ -0,0 +1,27 @@
+# Verify AppWrapper was dispatched and pod was created
+apiVersion: mcad.ibm.com/v1beta1
+kind: AppWrapper
+metadata:
+  name: deployment-silver-lo-pri-1replica
+  namespace: quota-errors
+  labels:
+    quota_context: "silver"
+    quota_service: "service-root"
+status:
+  state: Running
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: deployment-silver-lo-pri-1replica
+  namespace: quota-errors
+  labels:
+      app: deployment-silver-lo-pri-1replica
+      appwrapper.mcad.ibm.com: deployment-silver-lo-pri-1replica
+      resourceName: deployment-silver-lo-pri-1replica  
+status:
+    availableReplicas: 1
+    observedGeneration: 1
+    readyReplicas: 1
+    replicas: 1
+    updatedReplicas: 1