Deploying Redis Enterprise on Kubernetes

Quickstart Guide
Configuration
Upgrade
Supported K8S Distributions

This page describe how to deploy Redis Enterprise on Kubernetes using the Redis Enterprise Operator. High level architecture and overview of the solution can be found HERE.

Quickstart Guide

Prerequisites

A Kubernetes cluster version of 1.11 or higher, with a minimum of 3 worker nodes.
A Kubernetes client (kubectl) with a matching version. For OpenShift, an OpenShift client (oc).
Access to DockerHub, RedHat Container Catalog or a private repository that can serve the required images.

The following are the images and tags for this release:

Component	k8s	Openshift
Redis Enterprise	`redislabs/redis:6.0.20-69`	`redislabs/redis:6.0.20-69.rhel7-openshift`
Operator	`redislabs/operator:6.0.20-4`	`redislabs/operator:6.0.20-4`
Services Rigger	`redislabs/k8s-controller:6.0.20-4`	`redislabs/k8s-controller:6.0.20-4`

RedHat certified images are available on Redhat Catalog

Installation

The "Basic" installation deploys the operator (from the current release) from DockerHub and default settings. Recommended for KOPS, GKE, AKS, Rancher, VMWare Tanzu. This is the fastest way to get up and running with a new Redis Enterprise on Kubernetes.

Create a new namespace:

kubectl create namespace demo

Switch context to the newly created namespace:

kubectl config set-context --current --namespace=demo

Deploy the operator bundle

To deploy the default installation with kubectl, the following command will deploy a bundle of all the yaml declarations required for the operator:
```
kubectl apply -f bundle.yaml
```
Alternatively, to run each of the declarations of the bundle individually, run the following commands instead of the bundle:
```
kubectl apply -f role.yaml
kubectl apply -f role_binding.yaml
kubectl apply -f service_account.yaml
kubectl apply -f crds/v1/rec_crd.yaml
kubectl apply -f crds/v1alpha1/redb_crd.yaml
kubectl apply -f operator.yaml
```
Run kubectl get deployment and verify redis-enterprise-operator deployment is running.

A typical response may look like this:
```
NAME                               READY   UP-TO-DATE   AVAILABLE   AGE
redis-enterprise-operator          1/1     1            1           2m
```
Redis Enterprise Cluster custom resource - RedisEnterpriseCluster

Create a RedisEnterpriseCluster(REC) using the default configuration, which is suitable for development type deployments and works in typical scenarios. The full list of attributes supported through the Redis Enterprise Cluster (REC) API can be found HERE. Some examples can be found in the examples folder.
```
kubectl apply -f examples/v1/rec.yaml
```
Note: The Operator can only manage one Redis Enterprise Cluster custom resource in a namespace. To deploy another Enterprise Clusters in the same Kubernetes cluster, deploy an Operator in an additional namespace for each additional Enterprise Cluster required. Note that each Enterprise Cluster can effectively host hundreds of Redis Database instances. Deploying multiple clusters is typically used for scenarios where complete operational isolation is required at the cluster level.
Run kubectl get rec and verify creation was successful. rec is a shortcut for RedisEnterpriseCluster. The cluster takes around 5-10 minutes to come up. A typical response may look like this:
```
NAME  AGE
rec   5m
```
Note: Once the cluster is up, the cluster GUI and API could be used to configure databases. It is recommended to use the K8s REDB API that is configured through the following steps. To configure the cluster using the cluster GUI/API, use the ui service created by the operator and the default credentials as set in a secret. The secret name is the same as the cluster name within the namespace.

Redis Enterprise Database (REDB) Admission Controller: The Admission Controlller is recommended for use. It uses the Redis Enterprise Cluster to dynamically validate that REDB resources as configured by the operator are valid. Steps to configure the Admission Controller:

Install the Admission Controller via a bundle:

kubectl create -f admission.bundle.yaml

Wait for the secret to be created:

    kubectl get secret admission-tls
    NAME            TYPE     DATA   AGE
    admission-tls   Opaque   2      2m43s

Enable the Kubernetes webhook using the generated certificate

NOTE: One must replace REPLACE_WITH_NAMESPACE in the following command with the namespace the REC was installed into.

# save cert
CERT=`kubectl get secret admission-tls -o jsonpath='{.data.cert}'`
sed 's/NAMESPACE_OF_SERVICE_ACCOUNT/REPLACE_WITH_NAMESPACE/g' admission/webhook.yaml | kubectl create -f -

# create patch file
cat > modified-webhook.yaml <<EOF
webhooks:
- admissionReviewVersions:
  clientConfig:
    caBundle: $CERT
  name: redb.admission.redislabs
  admissionReviewVersions: ["v1beta1"]
EOF
# patch webhook with caBundle
kubectl patch ValidatingWebhookConfiguration redb-admission --patch "$(cat modified-webhook.yaml)"

Verify the installation In order to verify that the all the components of the Admission Controller are installed correctly, we will try to apply an invalid resource that should force the admission controller to reject it. If it applies succesfully, it means the admission controller has not been hooked up correctly.
```
$ kubectl apply -f - << EOF
apiVersion: app.redislabs.com/v1alpha1
kind: RedisEnterpriseDatabase
metadata:
  name: redis-enterprise-database
spec:
  evictionPolicy: illegal
EOF
```
This must fail with an error output by the admission webhook redb.admisison.redislabs that is being denied because it can't get the login credentials for the Redis Enterprise Cluster as none were specified.
```
Error from server: error when creating "STDIN": admission webhook "redb.admission.redislabs" denied the request: eviction_policy: u'illegal' is not one of [u'volatile-lru', u'volatile-ttl', u'volatile-random', u'allkeys-lru', u'allkeys-random', u'noeviction', u'volatile-lfu', u'allkeys-lfu']
```
Note: procedure to enable admission is documented with further detail [here](admission/README.md

Redis Enterprise Database custom resource - RedisEnterpriseDatabase

Create a RedisEnterpriseDatabase (REDB) by using Custom Resource.

Note: An example REDB.yaml file may be found [HERE] (https://github.com/RedisLabs/redis-enterprise-k8s-docs/tree/master/examples/v1alpha1). This is an alternative to copying and pasting the example code below into the CLI

The Redis Enterprise Operator can be instructed to manage databases on the Redis Enterprise Cluster using the REDB custom resource. Example:
```
cat << EOF > /tmp/redis-enterprise-database.yml
apiVersion: app.redislabs.com/v1alpha1
kind: RedisEnterpriseDatabase
metadata:
  name: redis-enterprise-database
spec:
  memorySize: 100MB
EOF
kubectl apply -f /tmp/redis-enterprise-database.yml
```
Replace the name of the cluster with the one used on the current namespace. All REDB configuration options are documented here.

Installation on OpenShift

The "OpenShift" installations deploys the operator from the current release with the RHEL image from DockerHub and default OpenShift settings. This is the fastest way to get up and running with a new cluster on OpenShift 3.x. For OpenShift 4.x, you may choose to use OLM deployment from within your OpenShift cluster or follow the steps below. Other custom configurations are referenced in this repository.

Note: you will need to replace <my-project> with your project name

Create a new project:
```
oc new-project my-project
```
Perform the following commands (you need cluster admin permissions for your Kubernetes cluster):
```
oc apply -f openshift/scc.yaml
```
You should receive the following response: securitycontextconstraints.security.openshift.io "redis-enterprise-scc" configured

Provide the operator permissions for pods (substitute your project for "my-project"):

oc adm policy add-scc-to-group redis-enterprise-scc system:serviceaccounts:my-project

Deploy the OpenShift operator bundle:

NOTE: Update the storageClassName setting in openshift.bundle.yaml (by default its set to gp2).
```
oc apply -f openshift.bundle.yaml
```
Redis Enterprise Cluster custom resource - RedisEnterpriseCluster

Apply the RedisEnterpriseCluster resource with RHEL7 based images:
```
oc apply -f openshift/rec_rhel.yaml
```

Install the Admission Controller via a bundle:

kubectl create -f admission.bundle.yaml

Wait for the secret to be created:

    kubectl get secret admission-tls
    NAME            TYPE     DATA   AGE
    admission-tls   Opaque   2      2m43s

Enable the Kubernetes webhook using the generated certificate

NOTE: One must replace REPLACE_WITH_NAMESPACE in the following command with the namespace the REC was installed into.

# save cert
CERT=`kubectl get secret admission-tls -o jsonpath='{.data.cert}'`
sed 's/NAMESPACE_OF_SERVICE_ACCOUNT/REPLACE_WITH_NAMESPACE/g' admission/webhook.yaml | kubectl create -f -

# create patch file
cat > modified-webhook.yaml <<EOF
webhooks:
- admissionReviewVersions:
  clientConfig:
    caBundle: $CERT
  name: redb.admission.redislabs
  admissionReviewVersions: ["v1beta1"]
EOF
# patch webhook with caBundle
kubectl patch ValidatingWebhookConfiguration redb-admission --patch "$(cat modified-webhook.yaml)"

Verify the installation In order to verify that the all the components of the Admission Controller are installed correctly, we will try to apply an invalid resource that should force the admission controller to reject it. If it applies succesfully, it means the admission controller has not been hooked up correctly.
```
$ kubectl apply -f - << EOF
apiVersion: app.redislabs.com/v1alpha1
kind: RedisEnterpriseDatabase
metadata:
  name: redis-enterprise-database
spec:
  evictionPolicy: illegal
EOF
```
This must fail with an error output by the admission webhook redb.admisison.redislabs that is being denied because it can't get the login credentials for the Redis Enterprise Cluster as none were specified.
```
Error from server: error when creating "STDIN": admission webhook "redb.admission.redislabs" denied the request: eviction_policy: u'illegal' is not one of [u'volatile-lru', u'volatile-ttl', u'volatile-random', u'allkeys-lru', u'allkeys-random', u'noeviction', u'volatile-lfu', u'allkeys-lfu']
```

Note: procedure to enable admission is documented with further detail [here](admission/README.md

Redis Enterprise Database custom resource - RedisEnterpriseDatabase

Create a RedisEnterpriseDatabase (REDB) by using Custom Resource. The Redis Enterprise Operator can be instructed to manage databases on the Redis Enterprise Cluster using the REDB custom resource. Example:
```
cat << EOF > /tmp/redis-enterprise-database.yml
apiVersion: app.redislabs.com/v1alpha1
kind: RedisEnterpriseDatabase
metadata:
  name: redis-enterprise-database
spec:

  memorySize: 100MB
EOF
kubectl apply -f /tmp/redis-enterprise-database.yml
```
Replace the name of the cluster with the one used on the current namespace. All REDB configuration options are documented here.

Installation on VMWare Tanzu

Instruction on how to deploy the Operator on PKS can be found on the Redis Labs documentation Website

Configuration

RedisEnterpriseCluster custom resource

The operator deploys a RedisEnterpriseCluster with default configurations values, but those can be customized in the RedisEnterpriseCluster spec as follow:

Redis Enterprise Image

  redisEnterpriseImageSpec:
    imagePullPolicy:  IfNotPresent
    repository:       redislabs/redis
    versionTag:       6.0.20-69

Persistence

  persistentSpec:
    enabled: true
    volumeSize: "10Gi" # if you don't provide default is 5 times RAM size
    storageClassName: "standard" #on AWS common storage class is gp2

Redis Enterprise Nodes(pods) resources

  redisEnterpriseNodeResources:
    limits:
      cpu: "4000m"
      memory: 4Gi
    requests:
      cpu: "4000m"
      memory: 4Gi

Cluster username (Default is demo@redislabs.com)
```
username: "admin@acme.com"
```

Extra Labels: Additional labels to tag the k8s resources created during deployment

  extraLabels:
    example1: "some-value"
    example2: "some-value"

UI service type: Load Balancer or cluster IP (default)
```
uiServiceType: LoadBalancer
```
Database service type (optional): Service types for access to databases. Should be a comma separated list. The possible values are cluster_ip, headless, and load_balancer. Default value is cluster_ip,headless. For example, to create a load_balancer type database service, explicitly add the following declaration to the Redis Enterprise Cluster spec:
```
servicesRiggerSpec:
  databaseServiceType: load_balancer
```

UI annotations: Add custom annotation to the UI service

  uiAnnotations:
    uiAnnotation1: 'UI-annotation1'
    uiAnnotation2: 'UI-Annotation2'

SideCar containers: images that will run along side the redis enterprise containers

  sideContainersSpec:
    - name: sidecar
      image: dockerhub_repo/repo:tag
      imagePullPolicy: IfNotPresent

CRDB (Active Active):

Currently supported for OpenShift*

activeActive: # edit values according to your cluster
  apiIngressUrl:  my-cluster1-api.myopenshiftcluster1.com
  dbIngressSuffix: -dbsuffix1.myopenshiftcluster1.com
  method: openShiftRoute

IPV4 enforcement

You might not have IPV6 support in your K8S cluster. In this case, you could enforce the use of IPV4, by adding the following attribute to the REC spec:
```
  enforceIPv4: true
```
Note: Setting 'enforceIPv4' to 'true' is a requirement for running REC on PKS.
Full detail can be found in Redis Enterprise Cluster Custom Resource Specification.

Private Repositories

Whenever images are not pulled from DockerHub, the following configuration must be specified:

In RedisEnterpriseClusterSpec (redis_enterprise_cluster.yaml):

redisEnterpriseImageSpec
redisEnterpriseServicesRiggerImageSpec
bootstrapperImageSpec

Image specifications in RedisEnterpriseClusterSpec follow the same schema:

see ImageSpec for full reference

For example:

  redisEnterpriseImageSpec:
    imagePullPolicy:  IfNotPresent
    repository:       harbor.corp.local/redisenterprise/redis
    versionTag:       6.0.20-69

  redisEnterpriseServicesRiggerImageSpec:
    imagePullPolicy:  IfNotPresent
    repository:       harbor.corp.local/redisenterprise/k8s-controller
    versionTag:       6.0.20-4

  bootstrapperImageSpec:
    imagePullPolicy:  IfNotPresent
    repository:       harbor.corp.local/redisenterprise/operator
    versionTag:       6.0.20-4

In Operator Deployment spec (operator.yaml):

For example:

spec:
  template:
    spec:
      containers:
        - name: redis-enterprise-operator
          image: harbor.corp.local/redisenterprise/operator:6.0.20-4

Image specification follow the K8s Container schema.

Pull secrets

Private repositories which require login can be accessed by creating a pull secret and declaring it in both the RedisEnterpriseClusterSpec and in the Operator Deployment spec.

Create a pull secret by running:

kubectl create secret docker-registry regcred --docker-server=<your-registry-server> --docker-username=<your-name> --docker-password=<your-pword> --docker-email=<your-email>

NOTE: Make sure to witch context to the REC namespace or add flag -n .

where:

<your-registry-server> is your Private repository FQDN. (https://index.docker.io/v1/ for DockerHub)
<your-name> is your Docker username.
<your-pword> is your Docker password.
<your-email> is your Docker email.

This creates a pull secret names regcred

To use in the RedisEnterpriseClusterSpec:

spec:
  pullSecrets:
    -name: regcred

To use in the Operator Deployment:

spec:
  template:
    spec:
      imagePullSecrets:
      -name: regcred

Advanced Configuration

To configure Priority Class, Node Pool, Eviction Thresholds and other advances configuration see topics.md file.
Full Redis Enterprise Cluster Custom Resource Specification
Full Redis Enterprise Database Custom Resource Specification

Upgrade

The Operator automates and simplifies the upgrade process.
The Redis Enterprise Cluster Software, and the Redis Enterprise Operator for Kubernetes versions are tightly coupled and should be upgraded together.
It is recommended to use the bundle.yaml to upgrade, as it loads all the relevant CRD documents for this version. If the updated CRDs are not loaded, the operator might fail. There are two ways to upgrade - either set 'autoUpgradeRedisEnterprise' within the Redis Enterprise Cluster Spec to instruct the operator to automatically upgrade to the compatible version, or specify the correct Redis Enterprise image manually using the versionTag attribute. The Redis Enterprise Version compatible with this release is 6.0.20-69

  autoUpgradeRedisEnterprise: true

Alternatively:

  RedisEnterpriseImageSpec:
    versionTag: redislabs/redis:6.0.20-69

Supported K8S Distributions

Each release of the Redis Enterprise Operator deployment is thoroughly tested against a set of Kubernetes distributions. The table below lists these, along with the current release's support status. "Supported", as well as "deprecated" support status indicates the current release has been tested in this environment and supported by RedisLabs. "Deprecated" also indicates that support will be dropped in a coming future release. "No longer supported" indicates that support has been dropped for this distribution. Any distribution that isn't explicitly listed is not supported for production workloads by RedisLabs. Supported versions (platforms/versions that are not listed are not supported):

Distribution	Support Status
Openshift 3.11 (K8s 1.11)	supported
Openshift 4.4 (K8s 1.17)	deprecated
OpenShift 4.5 (K8s 1.18)	supported
OpenShift 4.6 (K8s 1.19)	supported
OpenShift 4.7 (K8s 1.20)	supported
KOPS vanilla 1.15	deprecated
KOPS vanilla 1.16	supported
KOPS vanilla 1.17	supported
KOPS vanilla 1.18	supported
KOPS vanilla 1.19	supported
GKE 1.15	deprecated
GKE 1.16	deprecated
GKE 1.17	supported
GKE 1.18	supported
GKE 1.19	supported
Rancher 2.4 (K8s 1.17)	supported
Rancher 2.4 (K8s 1.18)	supported
Rancher 2.5 (K8s 1.17)	supported
Rancher 2.5 (K8s 1.18)	supported
Rancher 2.5 (K8s 1.19).	supported
VMWare TKGIE*** 1.7 (K8s 1.16)	supported
AKS 1.18	supported

* No longer supported by Red Hat ** No longer supported by Google *** Tanzu Kubernetes Grid Integrated Edition

quintonparker/redis-enterprise-k8s-docs