Chapter 2. Planning a deployment of AMQ Broker on OpenShift Container Platform
This section describes how to plan an Operator-based deployment.
Operators are programs that enable you to package, deploy, and manage OpenShift applications. Often, Operators automate common or complex tasks. Commonly, Operators are intended to provide:
- Consistent, repeatable installations
- Health checks of system components
- Over-the-air (OTA) updates
- Managed upgrades
Operators enable you to make changes while your broker instances are running, because they are always listening for changes to the Custom Resource (CR) instances that you used to configure your deployment. When you make changes to a CR, the Operator reconciles the changes with the existing broker deployment and updates the deployment to reflect the changes. In addition, the Operator provides a message migration capability, which ensures the integrity of messaging data. When a broker in a clustered deployment shuts down due to an intentional scaledown of the deployment, this capability migrates messages to a broker Pod that is still running in the same broker cluster.
2.1. Overview of the AMQ Broker Operator Custom Resource Definitions
In general, a Custom Resource Definition (CRD) is a schema of configuration items that you can modify for a custom OpenShift object deployed with an Operator. By creating a corresponding Custom Resource (CR) instance, you can specify values for configuration items in the CRD. If you are an Operator developer, what you expose through a CRD essentially becomes the API for how a deployed object is configured and used. You can directly access the CRD through regular HTTP curl
commands, because the CRD gets exposed automatically through Kubernetes.
You can install the AMQ Broker Operator using either the OpenShift command-line interface (CLI), or the Operator Lifecycle Manager, through the OperatorHub graphical interface. In either case, the AMQ Broker Operator includes the CRDs described below.
- Main broker CRD
You deploy a CR instance based on this CRD to create and configure a broker deployment.
Based on how you install the Operator, this CRD is:
-
The
broker_activemqartemis_crd
file in thecrds
directory of the Operator installation archive (OpenShift CLI installation method) -
The
ActiveMQArtemis
CRD in theCustom Resource Definitions
section of the OpenShift Container Platform web console (OperatorHub installation method)
-
The
- Address CRD
You deploy a CR instance based on this CRD to create addresses and queues for a broker deployment.
Based on how you install the Operator, this CRD is:
-
The
broker_activemqartemisaddress_crd
file in thecrds
directory of the Operator installation archive (OpenShift CLI installation method) -
The
ActiveMQArtemisAddresss
CRD in theCustom Resource Definitions
section of the OpenShift Container Platform web console (OperatorHub installation method)
-
The
- Security CRD
You deploy a CR instance based on this CRD to create users and associate those users with security contexts.
Based on how you install the Operator, this CRD is:
-
The
broker_activemqartemissecurity_crd
file in thecrds
directory of the Operator installation archive (OpenShift CLI installation method) -
The
ActiveMQArtemisSecurity
CRD in theCustom Resource Definitions
section of the OpenShift Container Platform web console (OperatorHub installation method).
-
The
- Scaledown CRD
The Operator automatically creates a CR instance based on this CRD when instantiating a scaledown controller for message migration.
Based on how you install the Operator, this CRD is:
-
The
broker_activemqartemisscaledown_crd
file in thecrds
directory of the Operator installation archive (OpenShift CLI installation method) -
The
ActiveMQArtemisScaledown
CRD in theCustom Resource Definitions
section of the OpenShift Container Platform web console (OperatorHub installation method).
-
The
Additional resources
To learn how to install the AMQ Broker Operator (and all included CRDs) using:
- The OpenShift CLI, see Section 3.2, “Installing the Operator using the CLI”
- The Operator Lifecycle Manager and OperatorHub graphical interface, see Section 3.3, “Installing the Operator using OperatorHub”.
For complete configuration references to use when creating CR instances based on the main broker and address CRDs, see:
2.2. Overview of the AMQ Broker Operator sample Custom Resources
The AMQ Broker Operator archive that you download and extract during installation includes sample Custom Resource (CR) files in the deploy/crs
directory. These sample CR files enable you to:
- Deploy a minimal broker without SSL or clustering.
- Define addresses.
The broker Operator archive that you download and extract also includes CRs for example deployments in the deploy/examples
directory, as listed below.
artemis-basic-deployment.yaml
- Basic broker deployment.
artemis-persistence-deployment.yaml
- Broker deployment with persistent storage.
artemis-cluster-deployment.yaml
- Deployment of clustered brokers.
artemis-persistence-cluster-deployment.yaml
- Deployment of clustered brokers with persistent storage.
artemis-ssl-deployment.yaml
- Broker deployment with SSL security.
artemis-ssl-persistence-deployment.yaml
- Broker deployment with SSL security and persistent storage.
artemis-aio-journal.yaml
- Use of asynchronous I/O (AIO) with the broker journal.
address-queue-create.yaml
- Address and queue creation.
2.3. Watch options for a Cluster Operator deployment
When the Cluster Operator is running, it starts to watch for updates of AMQ Broker custom resources (CRs).
You can choose to deploy the Cluster Operator to watch CRs from:
- A single namespace (the same namespace containing the Operator)
- All namespaces
If you have already installed a previous version of the AMQ Broker Operator in a namespace on your cluster, Red Hat recommends that you do not install the AMQ Broker Operator 7.10 version to watch that namespace to avoid potential conflicts.
2.4. How the Operator chooses container images
When you create a Custom Resource (CR) instance for a broker deployment, you do not need to explicitly specify broker or Init Container image names in the CR. By default, if you deploy a CR and do not explicitly specify container image values, the Operator automatically chooses the appropriate container images to use.
If you install the Operator using the OpenShift command-line interface, the Operator installation archive includes a sample CR file called broker_activemqartemis_cr.yaml
. In the sample CR, the spec.deploymentPlan.image
property is included and set to its default value of placeholder
. This value indicates that the Operator does not choose a broker container image until you deploy the CR.
The spec.deploymentPlan.initImage
property, which specifies the Init Container image, is not included in the broker_activemqartemis_cr.yaml
sample CR file. If you do not explicitly include the spec.deploymentPlan.initImage
property in your CR and specify a value, the Operator chooses an appropriate built-in Init Container image to use when you deploy the CR.
How the Operator chooses these images is described in this section.
To choose broker and Init Container images, the Operator first determines an AMQ Broker version to which the images should correspond. The Operator determines the version as follows:
-
If the
spec.upgrades.enabled
property in the main CR is already set totrue
and thespec.version
property specifies7.7.0
,7.8.0
,7.8.1
, or7.8.2
, the Operator uses that specified version. -
If
spec.upgrades.enabled
is not set totrue
, orspec.version
is set to an AMQ Broker version earlier than7.7.0
, the Operator uses the latest version of AMQ Broker (that is,7.10.7
).
The Operator then detects your container platform. The AMQ Broker Operator can run on the following container platforms:
- OpenShift Container Platform (x86_64)
- OpenShift Container Platform on IBM Z (s390x)
- OpenShift Container Platform on IBM Power Systems (ppc64le)
Based on the version of AMQ Broker and your container platform, the Operator then references two sets of environment variables in the operator.yaml
configuration file. These sets of environment variables specify broker and Init Container images for various versions of AMQ Broker, as described in the following sub-sections.
2.4.1. Environment variables for broker container images
The environment variables included in the operator.yaml
configuration file for broker container images have the following naming convention:
- OpenShift Container Platform
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Kubernetes_<AMQ_Broker_version_identifier>
- OpenShift Container Platform on IBM Z
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Kubernetes_<AMQ_Broker_version_identifier>_s390x
- OpenShift Container Platform on IBM Power Systems
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Kubernetes_<AMQ_Broker_version_identifier>_ppc64le
Environment variable names for each supported container platform and specific AMQ Broker versions are shown in the table.
Container platform | Environment variable names |
---|---|
OpenShift Container Platform |
|
OpenShift Container Platform on IBM Z |
|
OpenShift Container Platform on IBM Power Systems |
|
The value of each environment variable specifies a broker container image that is available from Red Hat. For example:
- name: RELATED_IMAGE_ActiveMQ_Artemis_Broker_Kubernetes_7100 #value: registry.redhat.io/amq7/amq-broker-rhel8:7.10 value: registry.redhat.io/amq7/amq-broker-rhel8@sha256:982ba18be1ac285722bc0ca8e85d2a42b8b844ab840b01425e79e3eeee6ee5b9
Therefore, based on an AMQ Broker version and your container platform, the Operator determines the applicable environment variable name. The Operator uses the corresponding image value when starting the broker container.
In the operator.yaml
file, the Operator uses an image that is represented by a Secure Hash Algorithm (SHA) value. The comment line, which begins with a number sign (#
) symbol, denotes that the SHA value corresponds to a specific container image tag.
2.4.2. Environment variables for Init Container images
The environment variables included in the operator.yaml
configuration file for Init Container images have the following naming convention:
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_<AMQ_Broker_version_identifier>
Environment variable names for specific AMQ Broker versions are listed below.
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_782
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_790
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_7100
The value of each environment variable specifies an Init Container image that is available from Red Hat. For example:
- name: RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_7100 #value: registry.redhat.io/amq7/amq-broker-init-rhel8:0.4-21 value: registry.redhat.io/amq7/amq-broker-init-rhel8@sha256:f37f98c809c6f29a83e3d5a3ac4494e28efe9b25d33c54f533c6a08662244622
Therefore, based on an AMQ Broker version, the Operator determines the applicable environment variable name. The Operator uses the corresponding image value when starting the Init Container.
As shown in the example, the Operator uses an image that is represented by a Secure Hash Algorithm (SHA) value. The comment line, which begins with a number sign (#
) symbol, denotes that the SHA value corresponds to a specific container image tag. Observe that the corresponding container image tag is not a floating tag in the form of 0.4-21
. This means that the container image used by the Operator remains fixed. The Operator does not automatically pull and use a new micro image version (that is, 0.4-21-n
, where n is the latest micro version) when it becomes available from Red Hat.
The environment variables included in the operator.yaml
configuration file for Init Container images have the following naming convention:
- OpenShift Container Platform
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_<AMQ_Broker_version_identifier>
- OpenShift Container Platform on IBM Z
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_s390x_<AMQ_Broker_version_identifier>
- OpenShift Container Platform on IBM Power Systems
-
RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_ppc64le_<AMQ_Broker_version_identifier>
Environment variable names for each supported container platform and specific AMQ Broker versions are shown in the table.
Container platform | Environment variable names |
---|---|
OpenShift Container Platform |
|
OpenShift Container Platform on IBM Z |
|
OpenShift Container Platform on IBM Power Systems |
|
The value of each environment variable specifies an Init Container image that is available from Red Hat. For example:
- name: RELATED_IMAGE_ActiveMQ_Artemis_Broker_Init_7100 #value: registry.redhat.io/amq7/amq-broker-init-rhel8:0.4-21-1 value: registry.redhat.io/amq7/amq-broker-init-rhel8@sha256:f37f98c809c6f29a83e3d5a3ac4494e28efe9b25d33c54f533c6a08662244622
Therefore, based on an AMQ Broker version and your container platform, the Operator determines the applicable environment variable name. The Operator uses the corresponding image value when starting the Init Container.
As shown in the example, the Operator uses an image that is represented by a Secure Hash Algorithm (SHA) value. The comment line, which begins with a number sign (#
) symbol, denotes that the SHA value corresponds to a specific container image tag. Observe that the corresponding container image tag is not a floating tag in the form of 0.4-21
. This means that the container image used by the Operator remains fixed. The Operator does not automatically pull and use a new micro image version (that is, 0.4-21-n
, where n is the latest micro version) when it becomes available from Red Hat.
Additional resources
- To learn how to use the AMQ Broker Operator to create a broker deployment, see Chapter 3, Deploying AMQ Broker on OpenShift Container Platform using the AMQ Broker Operator.
- For more information about how the Operator uses an Init Container to generate the broker configuration, see Section 4.1, “How the Operator generates the broker configuration”.
- To learn how to build and specify a custom Init Container image, see Section 4.7, “Specifying a custom Init Container image”.
2.5. Operator deployment notes
This section describes some important considerations when planning an Operator-based deployment
- Deploying the Custom Resource Definitions (CRDs) that accompany the AMQ Broker Operator requires cluster administrator privileges for your OpenShift cluster. When the Operator is deployed, non-administrator users can create broker instances via corresponding Custom Resources (CRs). To enable regular users to deploy CRs, the cluster administrator must first assign roles and permissions to the CRDs. For more information, see Creating cluster roles for Custom Resource Definitions in the OpenShift Container Platform documentation.
- When you update your cluster with the CRDs for the latest Operator version, this update affects all projects in the cluster. Any broker Pods deployed from previous versions of the Operator might become unable to update their status. When you click the Logs tab of a running broker Pod in the OpenShift Container Platform web console, you see messages indicating that 'UpdatePodStatus' has failed. However, the broker Pods and Operator in that project continue to work as expected. To fix this issue for an affected project, you must also upgrade that project to use the latest version of the Operator.
While you can create more than one broker deployment in a given OpenShift project by deploying multiple Custom Resource (CR) instances, typically, you create a single broker deployment in a project, and then deploy multiple CR instances for addresses.
Red Hat recommends you create broker deployments in separate projects.
If you intend to deploy brokers with persistent storage and do not have container-native storage in your OpenShift cluster, you need to manually provision Persistent Volumes (PVs) and ensure that these are available to be claimed by the Operator. For example, if you want to create a cluster of two brokers with persistent storage (that is, by setting
persistenceEnabled=true
in your CR), you need to have two persistent volumes available. By default, each broker instance requires storage of 2 GiB.If you specify
persistenceEnabled=false
in your CR, the deployed brokers uses ephemeral storage. Ephemeral storage means that that every time you restart the broker Pods, any existing data is lost.For more information about provisioning persistent storage in OpenShift Container Platform, see:
You must add configuration for the items listed below to the main broker CR instance before deploying the CR for the first time. You cannot add configuration for these items to a broker deployment that is already running.
-
If you update a parameter in your CR that the Operator is unable to dynamically update in the StatefulSet, the Operator deletes the StatefulSet and recreates it with the updated parameter value. Deleting the StatefulSet causes all pods to be deleted and recreated, which causes a temporary broker outage. An example of a CR update that the Operator cannot dynamically update in the StatefulSet is if you change
persistenceEnabled=false
topersistenceEnabled=true
.
2.6. Identifying namespaces watched by existing Operators
If the cluster already contains installed Operators for AMQ Broker, and you want a new Operator to watch all or multiple namespaces, you must ensure that the new Operator does not watch any of the same namespaces as existing Operators. Use the following procedure to identify the namespaces watched by existing Operators.
Procedure
-
In the left pane of the OpenShift Container Platform web console, click
. -
In the Project drop-down list, select
All Projects
. In the Filter Name box, specify a string, for example,
amq
, to display the Operators for AMQ Broker that are installed on the cluster.NoteThe namespace column displays the namespace where each operator is deployed.
Check the namespaces that each installed Operator for AMQ Broker is configured to watch.
- Click the Operator name to display the Operator details and click the YAML tab.
Search for
WATCH_NAMESPACE
and note the namespaces that the Operator watches.-
If the
WATCH_NAMESPACE
section has afieldPath
field that has a value ofmetadata.namespace
, the Operator is watching the namespace where it is deployed. If the
WATCH_NAMESPACE
section has avalue
field that has list of namespaces, the Operator is watching the specified namespaces. For example:- name: WATCH_NAMESPACE value: "namespace1, namespace2"
If the
WATCH_NAMESPACE
section has avalue
field that is empty or has an asterisk, the Operator is watching all the namespaces on the cluster. For example:- name: WATCH_NAMESPACE value: ""
In this case, before you deploy the new Operator, you must either uninstall the existing Operator or reconfigure it to watch specific namespaces.
-
If the
The procedures in the next section show you how to install the Operator and use Custom Resources (CRs) to create broker deployments on OpenShift Container Platform. When you have successfully completed the procedures, you will have the Operator running in an individual Pod. Each broker instance that you create will run as an individual Pod in a StatefulSet in the same project as the Operator. Later, you will you will see how to use a dedicated addressing CR to define addresses in your broker deployment.