github/flagger
1
0
Fork 0
mirror of https://github.com/fluxcd/flagger.git synced 2026-04-15 06:57:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add autoscaling support

Browse Source 
- add HorizontalPodAutoscaler reference to CRD
- create primary HPA on canary bootstrap
This commit is contained in:
Stefan Prodan
2018-10-11 11:16:56 +03:00
parent a03cf43a1d
commit a4f890c8b2
18 changed files with 180 additions and 83 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+54 -64
View File
@@ -45,25 +45,25 @@ Gated canary promotion stages:
* scan for canary deployments
* check Istio virtual service routes are mapped to primary and canary ClusterIP services
* check primary and canary deployments status
* halt rollout if a rolling update is underway
* halt rollout if pods are unhealthy
* halt advancement if a rolling update is underway
* halt advancement if pods are unhealthy
* increase canary traffic weight percentage from 0% to 5% (step weight)
* check canary HTTP request success rate and latency
* halt rollout if any metric is under the specified threshold
* halt advancement if any metric is under the specified threshold
* increment the failed checks counter
* check if the number of failed checks reached the threshold
* route all traffic to primary
* scale to zero the canary deployment and mark it as failed
* wait for the canary deployment to be updated (revision bump) and start over
* increase canary traffic weight by 5% (step weight) till it reaches 50% (max weight)
* halt rollout while canary request success rate is under the threshold
* halt rollout while canary request duration P99 is over the threshold
* halt rollout if the primary or canary deployment becomes unhealthy
* halt rollout while canary deployment is being scaled up/down by HPA
* halt advancement while canary request success rate is under the threshold
* halt advancement while canary request duration P99 is over the threshold
* halt advancement if the primary or canary deployment becomes unhealthy
* halt advancement while canary deployment is being scaled up/down by HPA
* promote canary to primary
* copy canary deployment spec template over primary
* wait for primary rolling update to finish
* halt rollout if pods are unhealthy
* halt advancement if pods are unhealthy
* route all traffic to primary
* scale to zero the canary deployment
* mark rollout as finished
@@ -80,19 +80,27 @@ metadata:
name: podinfo
namespace: test
spec:
# deployment reference
targetRef:
apiVersion: apps/v1
kind: Deployment
name: podinfo
# hpa reference (optional)
autoscalerRef:
apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
name: podinfo
service:
# container port
port: 9898
# Istio gateways (optional)
gateways:
- public-gateway.istio-system.svc.cluster.local
# Istio virtual service host names (optional)
hosts:
- podinfo.example.com
- app.istio.weavedx.com
canaryAnalysis:
# max number of failed metric checks
# before rolling back the canary
# max number of failed metric checks before rollback
threshold: 5
# max traffic percentage routed to canary
# percentage (0-100)
@@ -166,50 +174,68 @@ export REPO=https://raw.githubusercontent.com/stefanprodan/flagger/master
kubectl apply -f ${REPO}/artifacts/namespaces/test.yaml
```
Create a deployment:
Create a deployment and a horizontal pod autoscaler:
```bash
kubectl apply -f ${REPO}/artifacts/canaries/deployment.yaml
kubectl apply -f ${REPO}/artifacts/canaries/hpa.yaml
```
Create a canary promotion custom resource (replace the Istio gateway and the internet domain with your own)::
Create a canary promotion custom resource (replace the Istio gateway and the internet domain with your own):
```bash
kubectl apply -f ${REPO}/artifacts/canaries/canary.yaml
```
After a couple of seconds Flagger will create the canary objects:
```bash
canaries.flagger.app/podinfo
deployment.apps/podinfo
deployment.apps/podinfo-primary
horizontalpodautoscaler.autoscaling/podinfo
horizontalpodautoscaler.autoscaling/podinfo-primary
service/podinfo
service/podinfo-canary
service/podinfo-primary
virtualservices.networking.istio.io/podinfo
```
![flagger-canary-steps](https://raw.githubusercontent.com/stefanprodan/flagger/master/docs/diagrams/flagger-canary-steps.png)
Canary promotion output:
Trigger a canary deployment by updating the container image:
```bash
kubectl -n test set image deployment/podinfo \
podinfod=quay.io/stefanprodan/podinfo:1.2.1
```
Flagger detects that the deployment revision changed and starts a new rollout:
```
kubectl -n test describe canary/podinfo
Status:
Canary Revision: 16271121
Failed Checks: 6
Canary Revision: 19871136
Failed Checks: 0
State: finished
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Synced 3m flagger Starting canary deployment for podinfo.test
Normal Synced 3m flagger New revision detected podinfo.test
Normal Synced 3m flagger Scaling up podinfo.test
Warning Synced 3m flagger Waiting for podinfo.test rollout to finish: 0 of 1 updated replicas are available
Normal Synced 3m flagger Advance podinfo.test canary weight 5
Normal Synced 3m flagger Advance podinfo.test canary weight 10
Normal Synced 3m flagger Advance podinfo.test canary weight 15
Warning Synced 3m flagger Halt podinfo.test advancement request duration 2.525s > 500ms
Warning Synced 3m flagger Halt podinfo.test advancement request duration 1.567s > 500ms
Warning Synced 3m flagger Halt podinfo.test advancement request duration 823ms > 500ms
Normal Synced 2m flagger Advance podinfo.test canary weight 20
Normal Synced 2m flagger Advance podinfo.test canary weight 25
Normal Synced 1m flagger Advance podinfo.test canary weight 30
Warning Synced 1m flagger Halt podinfo.test advancement success rate 82.33% < 99%
Warning Synced 1m flagger Halt podinfo.test advancement success rate 87.22% < 99%
Warning Synced 1m flagger Halt podinfo.test advancement success rate 94.74% < 99%
Normal Synced 1m flagger Advance podinfo.test canary weight 35
Normal Synced 55s flagger Advance podinfo.test canary weight 40
Normal Synced 45s flagger Advance podinfo.test canary weight 45
Normal Synced 35s flagger Advance podinfo.test canary weight 50
Normal Synced 25s flagger Copying podinfo.test template spec to podinfo-primary.test
Warning Synced 15s flagger Waiting for podinfo.test rollout to finish: 1 of 2 updated replicas are available
Warning Synced 15s flagger Waiting for podinfo-primary.test rollout to finish: 1 of 2 updated replicas are available
Normal Synced 5s flagger Promotion completed! Scaling down podinfo.test
```
@@ -260,43 +286,6 @@ Events:
Warning Synced 1m flagger Canary failed! Scaling down podinfo.test
```
Trigger a new canary deployment by updating the canary image:
```bash
kubectl -n test set image deployment/podinfo \
podinfod=quay.io/stefanprodan/podinfo:1.2.1
```
Steer detects that the canary revision changed and starts a new rollout:
```
kubectl -n test describe canary/podinfo
Status:
Canary Revision: 19871136
Failed Checks: 0
State: finished
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Synced 3m flagger New revision detected podinfo-canary.test old 17211012 new 17246876
Normal Synced 3m flagger Scaling up podinfo.test
Warning Synced 3m flagger Waiting for podinfo.test rollout to finish: 0 of 1 updated replicas are available
Normal Synced 3m flagger Advance podinfo.test canary weight 5
Normal Synced 3m flagger Advance podinfo.test canary weight 10
Normal Synced 3m flagger Advance podinfo.test canary weight 15
Normal Synced 2m flagger Advance podinfo.test canary weight 20
Normal Synced 2m flagger Advance podinfo.test canary weight 25
Normal Synced 1m flagger Advance podinfo.test canary weight 30
Normal Synced 1m flagger Advance podinfo.test canary weight 35
Normal Synced 55s flagger Advance podinfo.test canary weight 40
Normal Synced 45s flagger Advance podinfo.test canary weight 45
Normal Synced 35s flagger Advance podinfo.test canary weight 50
Normal Synced 25s flagger Copying podinfo.test template spec to podinfo-primary.test
Warning Synced 15s flagger Waiting for podinfo.test rollout to finish: 1 of 2 updated replicas are available
Normal Synced 5s flagger Promotion completed! Scaling down podinfo.test
```
### Monitoring
Flagger comes with a Grafana dashboard made for canary analysis.
@@ -331,9 +320,10 @@ Advance podinfo.test canary weight 40
Halt podinfo.test advancement request duration 1.515s > 500ms
Advance podinfo.test canary weight 45
Advance podinfo.test canary weight 50
Copying podinfo-canary.test template spec to podinfo.test
Scaling down podinfo-canary.test
Promotion completed! podinfo-canary.test revision 81289
Copying podinfo.test template spec to podinfo-primary.test
Halt podinfo-primary.test advancement waiting for rollout to finish: 1 old replicas are pending termination
Scaling down podinfo.test
Promotion completed! podinfo.test
```
### Roadmap
artifacts/canaries/canary.yaml
+10 -2
View File
@@ -4,19 +4,27 @@ metadata:
name: podinfo
namespace: test
spec:
# deployment reference
targetRef:
apiVersion: apps/v1
kind: Deployment
name: podinfo
# HPA reference (optional)
autoscalerRef:
apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
name: podinfo
service:
# container port
port: 9898
# Istio gateways (optional)
gateways:
- public-gateway.istio-system.svc.cluster.local
# Istio virtual service host names (optional)
hosts:
- app.istio.weavedx.com
canaryAnalysis:
# max number of failed metric checks
# before rolling back the canary
# max number of failed metric checks before rollback
threshold: 5
# max traffic percentage routed to canary
# percentage (0-100)
artifacts/canaries/deployment.yaml
+2 -2
View File
@@ -22,7 +22,7 @@ spec:
spec:
containers:
- name: podinfod
image: quay.io/stefanprodan/podinfo:1.2.0
image: quay.io/stefanprodan/podinfo:1.2.1
imagePullPolicy: IfNotPresent
ports:
- containerPort: 9898
@@ -66,5 +66,5 @@ spec:
cpu: 2000m
memory: 512Mi
requests:
cpu: 10m
cpu: 100m
memory: 64Mi
artifacts/canaries/hpa.yaml
+19
View File
@@ -0,0 +1,19 @@
apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
name: podinfo
namespace: test
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: podinfo
minReplicas: 2
maxReplicas: 4
metrics:
- type: Resource
resource:
name: cpu
# scale up if usage is above
# 99% of the requested CPU (100m)
targetAverageUtilization: 99
artifacts/flagger/crd.yaml
+8
View File
@@ -31,6 +31,14 @@ spec:
type: string
name:
type: string
autoscalerRef:
properties:
apiVersion:
type: string
kind:
type: string
name:
type: string
service:
properties:
port:
artifacts/flagger/deployment.yaml
+1 -1
View File
@@ -22,7 +22,7 @@ spec:
serviceAccountName: flagger
containers:
- name: flagger
image: stefanprodan/flagger:0.0.1
image: stefanprodan/flagger:0.1.0-alpha.1
imagePullPolicy: Always
ports:
- name: http
charts/flagger/Chart.yaml
+2 -2
View File
@@ -1,6 +1,6 @@
apiVersion: v1
name: flagger
version: 0.0.1
appVersion: 0.0.1
version: 0.1.0
appVersion: 0.1.0-alpha.1
description: Flagger is a Kubernetes operator that automates the promotion of canary deployments using Istio routing for traffic shifting and Prometheus metrics for canary analysis.
home: https://github.com/stefanprodan/flagger
charts/flagger/templates/crd.yaml
+8
View File
@@ -32,6 +32,14 @@ spec:
type: string
name:
type: string
autoscalerRef:
properties:
apiVersion:
type: string
kind:
type: string
name:
type: string
service:
properties:
port:
charts/flagger/values.yaml
+1 -1
View File
@@ -2,7 +2,7 @@
image:
repository: stefanprodan/flagger
tag: 0.0.1
tag: 0.1.0-alpha.1
pullPolicy: IfNotPresent
controlLoopInterval: "10s"
docs/flagger-0.1.0.tgz
BIN
View File
Binary file not shown.
docs/grafana-5.2.4.tgz
BIN
View File
Binary file not shown.
docs/index.yaml
+18 -6
View File
@@ -1,9 +1,21 @@
apiVersion: v1
entries:
flagger:
- apiVersion: v1
appVersion: 0.1.0-alpha.1
created: 2018-10-11T11:16:00.30124+03:00
description: Flagger is a Kubernetes operator that automates the promotion of
canary deployments using Istio routing for traffic shifting and Prometheus metrics
for canary analysis.
digest: 3368bc9efbabe6043f399db46884f8a18319756b028ebe007ca69f9e81b72b3b
home: https://github.com/stefanprodan/flagger
name: flagger
urls:
- https://stefanprodan.github.io/flagger/flagger-0.1.0.tgz
version: 0.1.0
- apiVersion: v1
appVersion: 0.0.1
created: 2018-10-11T02:02:20.205592+03:00
created: 2018-10-11T11:16:00.300274+03:00
description: Flagger is a Kubernetes operator that automates the promotion of
canary deployments using Istio routing for traffic shifting and Prometheus metrics
for canary analysis.
@@ -16,10 +28,10 @@ entries:
grafana:
- apiVersion: v1
appVersion: 5.2.0
created: 2018-10-11T02:02:20.206046+03:00
created: 2018-10-11T11:16:00.302086+03:00
description: A Helm chart for monitoring progressive deployments powered by Istio
and Flagger
digest: 82e43b6777c39dfc6cf2dabf5d51148a48d64ed0efb3fa0cd09290602ccbfb87
digest: 109afd61ac5aaa212f18bfad4a1ad38425458f9d0a492547c5522d3898836faf
home: https://github.com/stefanprodan/flagger
name: grafana
urls:
@@ -28,9 +40,9 @@ entries:
podinfo-flagger:
- apiVersion: v1
appVersion: 1.2.1
created: 2018-10-11T02:02:20.206405+03:00
created: 2018-10-11T11:16:00.302902+03:00
description: Podinfo Helm chart for Flagger progressive delivery
digest: 2cddd159d311a1e20d08f486481c1743c70f9cecefc3ae7554b357a8186919d9
digest: 72d95f4f4ec1eea85ffc2a7cbb9b5f95bb56ea9d934ba24c5dde078d4f9c7605
engine: gotpl
home: https://github.com/stefanprodan/k8s-podinfo
maintainers:
@@ -58,4 +70,4 @@ entries:
urls:
- https://stefanprodan.github.io/flagger/podinfo-flagger-1.2.1.tgz
version: 1.2.1
generated: 2018-10-11T02:02:20.204687+03:00
generated: 2018-10-11T11:16:00.298073+03:00
docs/podinfo-flagger-2.0.0.tgz
BIN
View File
Binary file not shown.
pkg/apis/flagger/v1alpha1/types.go
+3
View File
@@ -40,6 +40,9 @@ type CanarySpec struct {
// reference to target resource
TargetRef hpav1.CrossVersionObjectReference `json:"targetRef"`
// reference to autoscaling resource
AutoscalerRef hpav1.CrossVersionObjectReference `json:"autoscalerRef"`
// virtual service spec
Service CanaryService `json:"service"`
pkg/apis/flagger/v1alpha1/zz_generated.deepcopy.go
+1
View File
@@ -152,6 +152,7 @@ func (in *CanaryService) DeepCopy() *CanaryService {
func (in *CanarySpec) DeepCopyInto(out *CanarySpec) {
*out = *in
out.TargetRef = in.TargetRef
out.AutoscalerRef = in.AutoscalerRef
in.Service.DeepCopyInto(&out.Service)
in.CanaryAnalysis.DeepCopyInto(&out.CanaryAnalysis)
return
pkg/controller/controller.go
+6 -4
View File
@@ -182,10 +182,12 @@ func (c *Controller) syncHandler(key string) error {
c.rollouts.Store(fmt.Sprintf("%s.%s", cd.Name, cd.Namespace), cd)
err = c.bootstrapDeployment(cd)
if err != nil {
c.logger.Warnf("%s.%s bootstrap error %v", cd.Name, cd.Namespace, err)
return err
if cd.Spec.TargetRef.Kind == "Deployment" {
err = c.bootstrapDeployment(cd)
if err != nil {
c.logger.Warnf("%s.%s bootstrap error %v", cd.Name, cd.Namespace, err)
return err
}
}
c.logger.Infof("Synced %s", key)
pkg/controller/deployer.go
+46
View File
@@ -6,6 +6,7 @@ import (
istiov1alpha3 "github.com/knative/pkg/apis/istio/v1alpha3"
flaggerv1 "github.com/stefanprodan/flagger/pkg/apis/flagger/v1alpha1"
appsv1 "k8s.io/api/apps/v1"
hpav1 "k8s.io/api/autoscaling/v2beta1"
corev1 "k8s.io/api/core/v1"
"k8s.io/apimachinery/pkg/api/errors"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -242,5 +243,50 @@ func (c *Controller) bootstrapDeployment(cd *flaggerv1.Canary) error {
c.recordEventInfof(cd, "VirtualService %s.%s created", virtualService.GetName(), cd.Namespace)
}
if cd.Spec.AutoscalerRef.Kind == "HorizontalPodAutoscaler" {
hpa, err := c.kubeClient.AutoscalingV2beta1().HorizontalPodAutoscalers(cd.Namespace).Get(cd.Spec.AutoscalerRef.Name, metav1.GetOptions{})
if err != nil {
if errors.IsNotFound(err) {
return fmt.Errorf("HorizontalPodAutoscaler %s.%s not found, retrying in %v",
cd.Spec.AutoscalerRef.Name, cd.Namespace, c.rolloutWindow)
} else {
return err
}
}
primaryHpaName := fmt.Sprintf("%s-primary", cd.Spec.AutoscalerRef.Name)
primaryHpa, err := c.kubeClient.AutoscalingV2beta1().HorizontalPodAutoscalers(cd.Namespace).Get(primaryHpaName, metav1.GetOptions{})
if errors.IsNotFound(err) {
primaryHpa = &hpav1.HorizontalPodAutoscaler{
ObjectMeta: metav1.ObjectMeta{
Name: primaryHpaName,
Namespace: cd.Namespace,
OwnerReferences: []metav1.OwnerReference{
*metav1.NewControllerRef(cd, schema.GroupVersionKind{
Group: flaggerv1.SchemeGroupVersion.Group,
Version: flaggerv1.SchemeGroupVersion.Version,
Kind: flaggerv1.CanaryKind,
}),
},
},
Spec: hpav1.HorizontalPodAutoscalerSpec{
ScaleTargetRef: hpav1.CrossVersionObjectReference{
Name: primaryName,
Kind: hpa.Spec.ScaleTargetRef.Kind,
APIVersion: hpa.Spec.ScaleTargetRef.APIVersion,
},
MinReplicas: hpa.Spec.MinReplicas,
MaxReplicas: hpa.Spec.MaxReplicas,
Metrics: hpa.Spec.Metrics,
},
}
_, err = c.kubeClient.AutoscalingV2beta1().HorizontalPodAutoscalers(cd.Namespace).Create(primaryHpa)
if err != nil {
return err
}
c.recordEventInfof(cd, "HorizontalPodAutoscaler %s.%s created", primaryHpa.GetName(), cd.Namespace)
}
}
return nil
}
pkg/version/version.go
+1 -1
View File
@@ -1,4 +1,4 @@
package version
var VERSION = "0.0.1"
var VERSION = "0.1.0-alpha.1"
var REVISION = "unknown"