Flagger Gateway for App Mesh is an Envoy-powered load balancer that exposes applications outside the mesh. The gateway facilitates canary deployments and A/B testing for user-facing web applications running on AWS App Mesh.

Prerequisites

Kubernetes >= 1.13
App Mesh controller >= 0.2.0
App Mesh inject >= 0.2.0

Installing the Chart

Add Flagger Helm repository:

$ helm repo add flagger https://flagger.app

Create a namespace with App Mesh sidecar injection enabled:

kubectl create ns flagger-system
kubectl label namespace test appmesh.k8s.aws/sidecarInjectorWebhook=enabled

Install App Mesh Gateway for an existing mesh:

helm upgrade -i appmesh-gateway flagger/appmesh-gateway \
--namespace flagger-system \
--set mesh.name=global

Optionally you can create a mesh at install time:

helm upgrade -i appmesh-gateway flagger/appmesh-gateway \
--namespace flagger-system \
--set mesh.name=global \
--set mesh.create=true

The configuration section lists the parameters that can be configured during installation.

Uninstalling the Chart

To uninstall/delete the appmesh-gateway deployment:

helm delete --purge appmesh-gateway

The command removes all the Kubernetes components associated with the chart and deletes the release.

Configuration

The following tables lists the configurable parameters of the chart and their default values.

Parameter	Description	Default
`service.type`	When set to LoadBalancer it creates an AWS NLB	`LoadBalancer`
`proxy.access_log_path`	to enable the access logs, set it to `/dev/stdout`	`/dev/null`
`proxy.image.repository`	image repository	`envoyproxy/envoy`
`proxy.image.tag`	image tag	`<VERSION>`
`proxy.image.pullPolicy`	image pull policy	`IfNotPresent`
`controller.image.repository`	image repository	`weaveworks/flagger-appmesh-gateway`
`controller.image.tag`	image tag	`<VERSION>`
`controller.image.pullPolicy`	image pull policy	`IfNotPresent`
`resources.requests/cpu`	pod CPU request	`100m`
`resources.requests/memory`	pod memory request	`128Mi`
`resources.limits/memory`	pod memory limit	`2Gi`
`nodeSelector`	node labels for pod assignment	`{}`
`tolerations`	list of node taints to tolerate	`[]`
`rbac.create`	if `true`, create and use RBAC resources	`true`
`rbac.pspEnabled`	If `true`, create and use a restricted pod security policy	`false`
`serviceAccount.create`	If `true`, create a new service account	`true`
`serviceAccount.name`	Service account to be used	None
`mesh.create`	If `true`, create mesh custom resource	`false`
`mesh.name`	The name of the mesh to use	`global`
`mesh.discovery`	The service discovery type to use, can be dns or cloudmap	`dns`
`hpa.enabled`	`true` if HPA resource should be created, metrics-server is required	`true`
`hpa.maxReplicas`	number of max replicas	`3`
`hpa.cpu`	average total CPU usage per pod (1-100)	`99`
`hpa.memory`	average memory usage per pod (100Mi-1Gi)	None
`discovery.optIn`	`true` if only services with the 'expose' annotation are discoverable	`true`