Skip to main content

Installing RabbitMQ Messaging Topology Operator

Overview

This guide covers the installation of the RabbitMQ Messaging Topology Operator in a Kubernetes cluster.

Compatibility

The Operator requires

  • Kubernetes 1.19 or above
  • RabbitMQ Cluster Operator 1.7.0+ (if not, RabbitMQ Messaging Topology Operator will fail to start)

Installation

There are two options for installing the Operator:

  1. Install with cert-manager
  2. Install generated certificates

Install with cert-manager

First, install cert-manager version 1.2.0+ on your cluster. For example, for version 1.3.1, run:

kubectl apply -f https://github.com/jetstack/cert-manager/releases/download/v1.3.1/cert-manager.yaml

Then, to install the Operator, run the following command:

kubectl apply -f https://github.com/rabbitmq/messaging-topology-operator/releases/latest/download/messaging-topology-operator-with-certmanager.yaml

Install with Generated Certificates

Without cert-manager installed, you will need to generate certificates used by admission webhooks yourself and include them in the operator and webhooks manifests.

Download the latest release manifest https://github.com/rabbitmq/messaging-topology-operator/releases/latest/download/messaging-topology-operator.yaml.

The Messaging Topology Operator has multiple admission webhooks. Their endpoints are TLS-enabled and require a webhook certificate that must be used in multiple places in the downloaded release manifest.

Sections below explain the steps involved into installing certificates for webhook admission.

Generate Key/Certificate Pair

First, generate one or more key/certificate pairs for webhook admission. These certificates must be valid for webhook-service.rabbitmq-system.svc.

Create a K8S Secret

Next, create a Kubernetes secret object with the name of webhook-server-cert in the rabbitmq-system namespace. The secret object must contain the following keys:

  • ca.crt (CA certificate)
  • tls.crt (leaf/webhook certificate)
  • tls.key (leaf/webhook private key)

The secret will be mounted to the Operator container, where all webhooks will run from.

For example:

apiVersion: v1
kind: Secret
type: kubernetes.io/tls
metadata:
name: webhook-server-cert
namespace: rabbitmq-system
data:
ca.crt: # ca cert that can be used to validate the webhook's server certificate
tls.crt: # generated certificate
tls.key: # generated private key

Use Generated Certificates in Release Manifest

Finally, add webhook's CA certificate to the release manifest, messaging-topology-operator.yaml. There are multiple admission webhooks, one for each CRD type.

Look for keyword clientConfig in the manifest, and paste the webhook CA certificate under clientConfig.caBundle. Because there are several webhooks, perform this action in several places.

The example below shows how to add a CA certificate to the queues.rabbitmq.com validating webhook:

apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
metadata:
annotations:
name: validating-webhook-configuration
webhooks:
- admissionReviewVersions:
- v1
clientConfig:
caBundle: # generated ca certificate goes in here
service:
name: webhook-service
namespace: rabbitmq-system
path: /validate-rabbitmq-com-v1beta1-queue
failurePolicy: Fail
name: vqueue.kb.io
rules:
- apiGroups:
- rabbitmq.com
...```


Then, to install the Operator, run the following command:

```bash
kubectl apply -f messaging-topology-operator.yaml

At this point, the RabbitMQ Messaging Topology Operator is successfully installed.

Using a non-default Kubernetes internal domain

By default, Kubernetes internal domain name is .cluster.local. This can be configured in kubeadm to be something else e.g. my.cluster.domain. In such cases, the Messaging Topology Operator can append the domain name to the connection strings it uses to interact with RabbitMQ.

To configure the Messaging Topology Operator to append the domain name in the connection string, set the environment variable MESSAGING_DOMAIN_NAME to your domain name e.g. ".my.cluster.domain".

To set this environment variable:

  • Download the installation manifest from the releases
  • Open the manifest and search for the Deployment with name messaging-topology-operator
  • Add a new element to the env list, with name MESSAGING_DOMAIN_NAME and value your domain name

The manifest related to the Deployment should look similar to this:

apiVersion: apps/v1
kind: Deployment
metadata:
[...]
name: messaging-topology-operator
namespace: rabbitmq-system
spec:
template:
[...]
spec:
containers:
- command:
- /manager
env:
- name: OPERATOR_NAMESPACE
valueFrom:
fieldRef:
fieldPath: metadata.namespace
- name: MESSAGING_DOMAIN_NAME
value: "my.cluster.domain"

Apply the edited manifest. Once the Pod has applied the changes, subsequent HTTP API requests to RabbitMQ will append the domain name to the connection string.

Modifying the RabbitMQ Messaging Topology Operator Default Configuration

This information describes how to modify the configuration of the RabbitMQ Messaging Topology Operator in a Kubernetes cluster.

To change the configuration, add or update the configuration environment variables by editing the Topology Operator deployment manifest. This is the YAML manifest artefact that is released with every new version of the RabbitMQ Cluster Operator on GitHub. The environment variables that can be set are listed in the table in Topology Operator Environment Variables.

Parameters

The following table listes the Topology Operator environment variables that are available to set in the deployment manifest.

Variable Name

Effect when Set

Effect when not Set

OPERATOR_SCOPE_NAMESPACE

Namespace, or list of namespaces, which the operator will reconcile and watch RabbitmqClusters (independent of installation namespace). Use a comma separator, without spaces e.g. "project-1,project-2,rabbitmq-testing"

All namespaces are watched and reconciled

ENABLE_DEBUG_PPROF

The default value is false because this variable should NOT be used in production. When it is set to true, it exposes a set of debug endpoints on the Operator Pod's metrics port for CPU and memory profiling of the Operator with pprof.

The pprof debug endpoint will not be exposed on the Operator Pod.

SYNC_PERIOD

Configure the operator to reconcile all owned objects periodically. It accepts string values with a time suffix e.g. "15m". It accepts any value parseable by time.ParseDuration function. By default, sync period is disabled, and reconciliation happens only when owned resources are updated.

Reconciliation will only happen when a resource is updated.

LEASE_DURATION

Time, in seconds, that non-leader candidates will wait to force acquire leadership. This is measured against time of last observed ack. Default is 15 seconds. The value must be a string e.g. "30".

Default value is set to 15 seconds.

RENEW_DEADLINE

Renew deadline is the duration that the acting controlplane will retry refreshing leadership before giving up. Default is 10 seconds. The value must be a string e.g. "10".

Default value is set to 10 seconds.

RETRY_PERIOD

Retry period is the duration the LeaderElector clients should wait between tries of actions. Default is 2 seconds. The value must be a string e.g. "3".

Default value is set to 2 seconds.

Older Operator Versions

To install a specific version of the Operator, obtain the manifest link from the Operator Releases. Using the latest version is strongly recommended.