1
0
mirror of https://github.com/openshift/openshift-docs.git synced 2026-02-05 12:46:18 +01:00
Files
openshift-docs/modules/nw-autoscaling-ingress-controller.adoc

222 lines
6.6 KiB
Plaintext

// Module included in the following assemblies:
//
// * networking/ingress-operator.adoc
:_mod-docs-content-type: PROCEDURE
[id="nw-autoscaling-ingress-controller_{context}"]
= Autoscaling an Ingress Controller
You can automatically scale an Ingress Controller to dynamically meet routing performance or availability requirements. For example, the requirement to increase throughput.
The following procedure provides an example for scaling up the default Ingress Controller.
.Prerequisites
* You have the {oc-first} installed.
* You have access to an {product-title} cluster as a user with the `cluster-admin` role.
* On {vmw-first}, bare-metal, and Nutanix installer-provisioned infrastructure, scaling up Ingress Controller pods does not improve external traffic performance. To improve performance, ensure that you complete the following prerequisites:
** You manually configured a user-managed load balancer for your cluster.
** You ensured that the load balancer was configured for the cluster nodes that handle incoming traffic from the Ingress Controller.
* You installed the Custom Metrics Autoscaler Operator and an associated KEDA Controller.
** You can install the Operator by using the software catalog on the web console. After you install the Operator, you can create an instance of `KedaController`.
.Procedure
. Create a service account to authenticate with Thanos by running the following command:
+
[source,terminal]
----
$ oc create -n openshift-ingress-operator serviceaccount thanos && oc describe -n openshift-ingress-operator serviceaccount thanos
----
+
.Example output
[source,terminal]
----
Name: thanos
Namespace: openshift-ingress-operator
Labels: <none>
Annotations: <none>
Image pull secrets: thanos-dockercfg-kfvf2
Mountable secrets: thanos-dockercfg-kfvf2
Tokens: <none>
Events: <none>
----
. Manually create the service account secret token with the following command:
+
[source,terminal]
----
$ oc apply -f - <<EOF
apiVersion: v1
kind: Secret
metadata:
name: thanos-token
namespace: openshift-ingress-operator
annotations:
kubernetes.io/service-account.name: thanos
type: kubernetes.io/service-account-token
EOF
----
. Define a `TriggerAuthentication` object within the `openshift-ingress-operator` namespace by using the service account's token.
.. Create the `TriggerAuthentication` object and pass the value of the `secret` variable to the `TOKEN` parameter:
+
[source,terminal]
----
$ oc apply -f - <<EOF
apiVersion: keda.sh/v1alpha1
kind: TriggerAuthentication
metadata:
name: keda-trigger-auth-prometheus
namespace: openshift-ingress-operator
spec:
secretTargetRef:
- parameter: bearerToken
name: thanos-token
key: token
- parameter: ca
name: thanos-token
key: ca.crt
EOF
----
. Create and apply a role for reading metrics from Thanos:
.. Create a new role, `thanos-metrics-reader.yaml`, that reads metrics from pods and nodes:
+
.thanos-metrics-reader.yaml
[source,yaml]
----
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
name: thanos-metrics-reader
namespace: openshift-ingress-operator
rules:
- apiGroups:
- ""
resources:
- pods
- nodes
verbs:
- get
- apiGroups:
- metrics.k8s.io
resources:
- pods
- nodes
verbs:
- get
- list
- watch
- apiGroups:
- ""
resources:
- namespaces
verbs:
- get
----
.. Apply the new role by running the following command:
+
[source,terminal]
----
$ oc apply -f thanos-metrics-reader.yaml
----
. Add the new role to the service account by entering the following commands:
+
[source,terminal]
----
$ oc adm policy -n openshift-ingress-operator add-role-to-user thanos-metrics-reader -z thanos --role-namespace=openshift-ingress-operator
----
+
[source,terminal]
----
$ oc adm policy -n openshift-ingress-operator add-cluster-role-to-user cluster-monitoring-view -z thanos
----
+
[NOTE]
====
The argument `add-cluster-role-to-user` is only required if you use cross-namespace queries. The following step uses a query from the `kube-metrics` namespace which requires this argument.
====
. Create a new `ScaledObject` YAML file, `ingress-autoscaler.yaml`, that targets the default Ingress Controller deployment:
+
.Example `ScaledObject` definition
[source,yaml]
----
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
name: ingress-scaler
namespace: openshift-ingress-operator
spec:
scaleTargetRef: <1>
apiVersion: operator.openshift.io/v1
kind: IngressController
name: default
envSourceContainerName: ingress-operator
minReplicaCount: 1
maxReplicaCount: 20 <2>
cooldownPeriod: 1
pollingInterval: 1
triggers:
- type: prometheus
metricType: AverageValue
metadata:
serverAddress: https://thanos-querier.openshift-monitoring.svc.cluster.local:9091 <3>
namespace: openshift-ingress-operator <4>
metricName: 'kube-node-role'
threshold: '1'
query: 'sum(kube_node_role{role="worker",service="kube-state-metrics"})' <5>
authModes: "bearer"
authenticationRef:
name: keda-trigger-auth-prometheus
----
<1> The custom resource that you are targeting. In this case, the Ingress Controller.
<2> Optional: The maximum number of replicas. If you omit this field, the default maximum is set to 100 replicas.
<3> The Thanos service endpoint in the `openshift-monitoring` namespace.
<4> The Ingress Operator namespace.
<5> This expression evaluates to however many worker nodes are present in the deployed cluster.
+
[IMPORTANT]
====
If you are using cross-namespace queries, you must target port 9091 and not port 9092 in the `serverAddress` field. You also must have elevated privileges to read metrics from this port.
====
. Apply the custom resource definition by running the following command:
+
[source,terminal]
----
$ oc apply -f ingress-autoscaler.yaml
----
.Verification
* Verify that the default Ingress Controller is scaled out to match the value returned by the `kube-state-metrics` query by running the following commands:
** Use the `grep` command to search the Ingress Controller YAML file for the number of replicas:
+
[source,terminal]
----
$ oc get -n openshift-ingress-operator ingresscontroller/default -o yaml | grep replicas:
----
** Get the pods in the `openshift-ingress` project:
+
[source,terminal]
----
$ oc get pods -n openshift-ingress
----
+
.Example output
[source,terminal]
----
NAME READY STATUS RESTARTS AGE
router-default-7b5df44ff-l9pmm 2/2 Running 0 17h
router-default-7b5df44ff-s5sl5 2/2 Running 0 3d22h
router-default-7b5df44ff-wwsth 2/2 Running 0 66s
----