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.
- 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:6.0.6-39 |
redislabs/redis:6.0.6-39.rhel7-openshift |
Operator | redislabs/operator:6.0.6-11 |
redislabs/operator:6.0.6-11.rhel7 |
Services Rigger | redislabs/k8s-controller:6.0.6-11 |
redislabs/k8s-controller:6.0.6-11.rhel7 |
- RedHat certified images are available on Redhat Catalog
The "Basic" installation deploys the operator (from the current release) with the default Ubuntu/Alpine base OS images from DockerHub and default settings. 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/app_v1_redisenterprisecluster_crd.yaml kubectl apply -f crds/app_v1alpha1_redisenterprisedatabase_crd.yaml kubectl apply -f operator.yaml
Note: The rbac.yaml file used in previous releases has been broken down into three distinct files:
role.yaml
,role_binding.yaml
andservice_account.yaml
. Thecrd.yaml
file was renamed toredisenterprisecluster_crd.yaml
, with the API version prepended to the filename. -
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. For more advanced deployment options you may choose the configuration relevant for you - see the index at the top for documentation references that cover many scenarios and the examples in the example folder.kubectl apply -f crds/app_v1_redisenterprisecluster_cr.yaml
Note: The redis-enterprise-cluster.yaml file was renamed to redisenterprisecluster_cr.yaml, with the API version prepended to the filename.
-
Run
kubectl get rec
and verify creation was successful.rec
is a shortcut for RedisEnterpriseCluster. A typical response may look like this:NAME AGE redis-enterprise 5m
-
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: redisEnterpriseCluster: name: redis-enterprise 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.
Optional: REDB admission controller
When using the REDB Custom Resource Definition (Redis Enterprise Database) it is recommended to set up admission control to improve input validation and catch configuration errors before they reach the cluster. The procedure is documented here
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 inopenshift.bundle.yaml
(by default its set togp2
).oc apply -f openshift.bundle.yaml
-
Redis Enterprise Cluster custom resource -
RedisEnterpriseCluster
Apply the
RedisEnterpriseCluster
resource with RHEL7 based images:kubectl apply -f openshift/redis-enterprise-cluster_rhel.yaml
-
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: redisEnterpriseCluster: name: redis-enterprise 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.
Optional: REDB admission controller
When using the REDB Custom Resource Definition (Redis Enterprise Database) it is recommended to set up admission controller to improve input validation and catch configuration errors before they reach the cluster. The procedure is documented here.
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: 6.0.6-39
-
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 [email protected])
username: "[email protected]"
-
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
-
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.
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: 6.0.6-39
redisEnterpriseServicesRiggerImageSpec:
imagePullPolicy: IfNotPresent
repository: harbor.corp.local/redisenterprise/k8s-controller
versionTag: 6.0.6-11
bootstrapperImageSpec:
imagePullPolicy: IfNotPresent
repository: harbor.corp.local/redisenterprise/operator
versionTag: 6.0.6-11
In Operator Deployment spec (operator.yaml):
For example:
spec:
template:
spec:
containers:
- name: redis-enterprise-operator
image: harbor.corp.local/redisenterprise/operator:6.0.6-11
Image specification follow the K8s Container schema.
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>
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
- 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
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 6.0.6-39
autoUpgradeRedisEnterprise: true
Alternatively:
RedisEnterpriseImageSpec:
versionTag: redislabs/redis:6.0.6-39