emansije/kube-metrics-adapter
1
0
Fork 0
forked from yusheng-guo/kube-metrics-adapter
Code Pull Requests Activity

Added weighting of RPS metrics based on backend weights (#27)

Browse Source 
* Added weighting of rps metrics based on backend weights

Signed-off-by: Arjun Naik <arjun.rn@gmail.com>

* Updated documented with instructions on how to use the backend weighting

Signed-off-by: Arjun Naik <arjun.rn@gmail.com>

* Changed separator for RPS metric and added flag to specify backend weights annotation.

Signed-off-by: Arjun Naik <arjun.rn@gmail.com>

* Allow for multiple backends with for weighting.

Signed-off-by: Arjun Naik <arjun.rn@gmail.com>
This commit is contained in:
Arjun
2019-01-17 13:13:52 +01:00
committed by GitHub
parent f49f7821dc
commit 72aa672f51
5 changed files with 301 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

189
pkg/collector/skipper_collector_test.go
View File

@@ -1,7 +1,16 @@
package collector
import (
"encoding/json"
"fmt"
"github.com/stretchr/testify/assert"
corev1 "k8s.io/api/core/v1"
"k8s.io/api/extensions/v1beta1"
"k8s.io/apimachinery/pkg/api/resource"
"k8s.io/client-go/kubernetes"
"k8s.io/metrics/pkg/apis/custom_metrics"
"testing"
"time"
"github.com/stretchr/testify/require"
"k8s.io/api/apps/v1"
@@ -10,11 +19,16 @@ import (
"k8s.io/client-go/kubernetes/fake"
)
const (
testBackendWeightsAnnotation = "zalando.org/backend-weights"
testStacksetWeightsAnnotation = "zalando.org/stack-set-weights"
)
func TestTargetRefReplicasDeployments(t *testing.T) {
client := fake.NewSimpleClientset()
name := "some-app"
defaultNamespace := "default"
deployment, err := newDeployment(client, defaultNamespace, name)
deployment, err := newDeployment(client, defaultNamespace, name, 2, 1)
require.NoError(t, err)
// Create an HPA with the deployment as ref
@@ -59,7 +73,7 @@ func newHPA(namesapce string, refName string, refKind string) *autoscalingv2beta
}
}
func newDeployment(client *fake.Clientset, namespace string, name string) (*v1.Deployment, error) {
func newDeployment(client *fake.Clientset, namespace string, name string, replicas, readyReplicas int32) (*v1.Deployment, error) {
return client.AppsV1().Deployments(namespace).Create(&v1.Deployment{
ObjectMeta: metav1.ObjectMeta{
Name: name,
@@ -67,8 +81,8 @@ func newDeployment(client *fake.Clientset, namespace string, name string) (*v1.D
},
Spec: v1.DeploymentSpec{},
Status: v1.DeploymentStatus{
ReadyReplicas: 1,
Replicas: 2,
ReadyReplicas: replicas,
Replicas: readyReplicas,
},
})
}
@@ -85,3 +99,170 @@ func newStatefulSet(client *fake.Clientset, namespace string, name string) (*v1.
},
})
}
func TestSkipperCollector(t *testing.T) {
for _, tc := range []struct {
msg string
metrics []int
backend string
ingressName string
collectedMetric int
namespace string
backendWeights map[string]map[string]int
replicas int32
readyReplicas int32
backendAnnotations []string
}{
{
msg: "test unweighted hpa",
metrics: []int{1000, 1000, 2000},
ingressName: "dummy-ingress",
collectedMetric: 2000,
namespace: "default",
backend: "dummy-backend",
replicas: 1,
readyReplicas: 1,
},
{
msg: "test weighted backend",
metrics: []int{100, 1500, 700},
ingressName: "dummy-ingress",
collectedMetric: 600,
namespace: "default",
backend: "backend1",
backendWeights: map[string]map[string]int{testBackendWeightsAnnotation: {"backend2": 60, "backend1": 40}},
replicas: 1,
readyReplicas: 1,
backendAnnotations: []string{testBackendWeightsAnnotation},
},
{
msg: "test multiple replicas",
metrics: []int{100, 1500, 700},
ingressName: "dummy-ingress",
collectedMetric: 150,
namespace: "default",
backend: "backend1",
backendWeights: map[string]map[string]int{testBackendWeightsAnnotation: {"backend2": 50, "backend1": 50}},
replicas: 5,
readyReplicas: 5,
backendAnnotations: []string{testBackendWeightsAnnotation},
},
{
msg: "test multiple backend annotation",
metrics: []int{100, 1500, 700},
ingressName: "dummy-ingress",
collectedMetric: 300,
namespace: "default",
backend: "backend1",
backendWeights: map[string]map[string]int{
testBackendWeightsAnnotation: {"backend2": 20, "backend1": 80},
testStacksetWeightsAnnotation: {"backend2": 0, "backend1": 100},
},
replicas: 5,
readyReplicas: 5,
backendAnnotations: []string{testBackendWeightsAnnotation, testStacksetWeightsAnnotation},
},
} {
t.Run(tc.msg, func(tt *testing.T) {
client := fake.NewSimpleClientset()
makeIngress(client, tc.namespace, tc.ingressName, tc.backend, tc.backendWeights)
plugin := makePlugin(tc.metrics)
hpa := makeHPA(tc.ingressName, tc.backend)
config := makeConfig(tc.backend)
newDeployment(client, tc.namespace, tc.backend, tc.replicas, tc.readyReplicas)
collector, err := NewSkipperCollector(client, plugin, hpa, config, time.Minute, tc.backendAnnotations, tc.backend)
assert.NoError(tt, err, "failed to create skipper collector: %v", err)
collected, err := collector.GetMetrics()
assert.NoError(tt, err, "failed to collect metrics: %v", err)
assert.Len(t, collected, 1, "the number of metrics returned is not 1")
assert.EqualValues(t, tc.collectedMetric, collected[0].Custom.Value.Value(), "the returned metric is not expected value")
})
}
}
func makeIngress(client kubernetes.Interface, namespace, ingressName, backend string, backendWeights map[string]map[string]int) error {
annotations := make(map[string]string)
for anno, weights := range backendWeights {
sWeights, err := json.Marshal(weights)
if err != nil {
return err
}
annotations[anno] = string(sWeights)
}
_, err := client.ExtensionsV1beta1().Ingresses(namespace).Create(&v1beta1.Ingress{
ObjectMeta: metav1.ObjectMeta{
Name: ingressName,
Annotations: annotations,
},
Spec: v1beta1.IngressSpec{
Backend: &v1beta1.IngressBackend{
ServiceName: backend,
},
TLS: nil,
Rules: []v1beta1.IngressRule{
{
Host: "example.org",
},
},
},
Status: v1beta1.IngressStatus{
LoadBalancer: corev1.LoadBalancerStatus{
Ingress: nil,
},
},
})
return err
}
func makeHPA(ingressName, backend string) *autoscalingv2beta1.HorizontalPodAutoscaler {
return &autoscalingv2beta1.HorizontalPodAutoscaler{
Spec: autoscalingv2beta1.HorizontalPodAutoscalerSpec{
ScaleTargetRef: autoscalingv2beta1.CrossVersionObjectReference{
Kind: "Deployment",
Name: backend,
},
Metrics: []autoscalingv2beta1.MetricSpec{
{
Type: autoscalingv2beta1.ObjectMetricSourceType,
Object: &autoscalingv2beta1.ObjectMetricSource{
Target: autoscalingv2beta1.CrossVersionObjectReference{Name: ingressName, APIVersion: "extensions/v1", Kind: "Ingress"},
MetricName: fmt.Sprintf("%s,%s", rpsMetricName, backend),
},
},
},
},
}
}
func makeConfig(backend string) *MetricConfig {
return &MetricConfig{
MetricTypeName: MetricTypeName{Name: fmt.Sprintf("%s,%s", rpsMetricName, backend)},
}
}
type FakeCollectorPlugin struct {
metrics []CollectedMetric
}
type FakeCollector struct {
metrics []CollectedMetric
}
func (c *FakeCollector) GetMetrics() ([]CollectedMetric, error) {
return c.metrics, nil
}
func (FakeCollector) Interval() time.Duration {
return time.Minute
}
func (p *FakeCollectorPlugin) NewCollector(hpa *autoscalingv2beta1.HorizontalPodAutoscaler, config *MetricConfig, interval time.Duration) (Collector, error) {
return &FakeCollector{metrics: p.metrics}, nil
}
func makePlugin(metrics []int) CollectorPlugin {
m := make([]CollectedMetric, len(metrics))
for i, v := range metrics {
m[i] = CollectedMetric{Custom: custom_metrics.MetricValue{Value: *resource.NewQuantity(int64(v), resource.DecimalSI)}}
}
return &FakeCollectorPlugin{metrics: m}
}