Images
Creating and managing images and imagestreams in OpenShift Container Platform
Abstract
Chapter 1. Overview of images
1.1. Understanding containers, images, and image streams
Containers, images, and image streams are important concepts to understand when you set out to create and manage containerized software. An image holds a set of software that is ready to run, while a container is a running instance of a container image. An image stream provides a way of storing different versions of the same basic image. Those different versions are represented by different tags on the same image name.
1.2. Images
Containers in OpenShift Container Platform are based on OCI- or Docker-formatted container images. An image is a binary that includes all of the requirements for running a single container, as well as metadata describing its needs and capabilities.
You can think of it as a packaging technology. Containers only have access to resources defined in the image unless you give the container additional access when creating it. By deploying the same image in multiple containers across multiple hosts and load balancing between them, OpenShift Container Platform can provide redundancy and horizontal scaling for a service packaged into an image.
You can use the podman or docker
CLI directly to build images, but OpenShift Container Platform also supplies builder images that assist with creating new images by adding your code or configuration to existing images.
Because applications develop over time, a single image name can actually refer to many different versions of the same image. Each different image is referred to uniquely by its hash, a long hexadecimal number such as fd44297e2ddb050ec4f…
, which is usually shortened to 12 characters, such as fd44297e2ddb
.
1.3. Image registry
An image registry is a content server that can store and serve container images. For example:
registry.redhat.io
A registry contains a collection of one or more image repositories, which contain one or more tagged images. Red Hat provides a registry at registry.redhat.io
for subscribers. OpenShift Container Platform can also supply its own OpenShift image registry for managing custom container images.
1.4. Image repository
An image repository is a collection of related container images and tags identifying them. For example, the OpenShift Container Platform Jenkins images are in the repository:
docker.io/openshift/jenkins-2-centos7
1.5. Image tags
An image tag is a label applied to a container image in a repository that distinguishes a specific image from other images in an image stream. Typically, the tag represents a version number of some sort. For example, here :v3.11.59-2
is the tag:
registry.access.redhat.com/openshift3/jenkins-2-rhel7:v3.11.59-2
You can add additional tags to an image. For example, an image might be assigned the tags :v3.11.59-2
and :latest
.
OpenShift Container Platform provides the oc tag
command, which is similar to the docker tag
command, but operates on image streams instead of directly on images.
1.6. Image IDs
An image ID is a SHA (Secure Hash Algorithm) code that can be used to pull an image. A SHA image ID cannot change. A specific SHA identifier always references the exact same container image content. For example:
docker.io/openshift/jenkins-2-centos7@sha256:ab312bda324
1.7. Containers
The basic units of OpenShift Container Platform applications are called containers. Linux container technologies are lightweight mechanisms for isolating running processes so that they are limited to interacting with only their designated resources. The word container is defined as a specific running or paused instance of a container image.
Many application instances can be running in containers on a single host without visibility into each others' processes, files, network, and so on. Typically, each container provides a single service, often called a micro-service, such as a web server or a database, though containers can be used for arbitrary workloads.
The Linux kernel has been incorporating capabilities for container technologies for years. The Docker project developed a convenient management interface for Linux containers on a host. More recently, the Open Container Initiative has developed open standards for container formats and container runtimes. OpenShift Container Platform and Kubernetes add the ability to orchestrate OCI- and Docker-formatted containers across multi-host installations.
Though you do not directly interact with container runtimes when using OpenShift Container Platform, understanding their capabilities and terminology is important for understanding their role in OpenShift Container Platform and how your applications function inside of containers.
Tools such as podman can be used to replace docker
command-line tools for running and managing containers directly. Using podman
, you can experiment with containers separately from OpenShift Container Platform.
1.8. Why use imagestreams
An image stream and its associated tags provide an abstraction for referencing container images from within OpenShift Container Platform. The image stream and its tags allow you to see what images are available and ensure that you are using the specific image you need even if the image in the repository changes.
Image streams do not contain actual image data, but present a single virtual view of related images, similar to an image repository.
You can configure builds and deployments to watch an image stream for notifications when new images are added and react by performing a build or deployment, respectively.
For example, if a deployment is using a certain image and a new version of that image is created, a deployment could be automatically performed to pick up the new version of the image.
However, if the image stream tag used by the deployment or build is not updated, then even if the container image in the container image registry is updated, the build or deployment continues using the previous, presumably known good image.
The source images can be stored in any of the following:
- OpenShift Container Platform’s integrated registry.
- An external registry, for example registry.redhat.io or quay.io.
- Other image streams in the OpenShift Container Platform cluster.
When you define an object that references an image stream tag, such as a build or deployment configuration, you point to an image stream tag and not the repository. When you build or deploy your application, OpenShift Container Platform queries the repository using the image stream tag to locate the associated ID of the image and uses that exact image.
The image stream metadata is stored in the etcd instance along with other cluster information.
Using image streams has several significant benefits:
- You can tag, rollback a tag, and quickly deal with images, without having to re-push using the command line.
- You can trigger builds and deployments when a new image is pushed to the registry. Also, OpenShift Container Platform has generic triggers for other resources, such as Kubernetes objects.
- You can mark a tag for periodic re-import. If the source image has changed, that change is picked up and reflected in the image stream, which triggers the build or deployment flow, depending upon the build or deployment configuration.
- You can share images using fine-grained access control and quickly distribute images across your teams.
- If the source image changes, the image stream tag still points to a known-good version of the image, ensuring that your application does not break unexpectedly.
- You can configure security around who can view and use the images through permissions on the image stream objects.
- Users that lack permission to read or list images on the cluster level can still retrieve the images tagged in a project using image streams.
You can manage image streams, use image streams with Kubernetes resources, and trigger updates on image stream updates.
1.9. Image stream tags
An image stream tag is a named pointer to an image in an image stream. An image stream tag is similar to a container image tag.
1.10. Image stream images
An image stream image allows you to retrieve a specific container image from a particular image stream where it is tagged. An image stream image is an API resource object that pulls together some metadata about a particular image SHA identifier.
1.11. Image stream triggers
An image stream trigger causes a specific action when an image stream tag changes. For example, importing can cause the value of the tag to change, which causes a trigger to fire when there are deployments, builds, or other resources listening for those.
1.12. How you can use the Cluster Samples Operator
During the initial startup, the Operator creates the default samples resource to initiate the creation of the image streams and templates. You can use the Cluster Samples Operator to manage the sample image streams and templates stored in the openshift
namespace.
As a cluster administrator, you can use the Cluster Samples Operator to:
1.13. About templates
A template is a definition of an object to be replicated. You can use templates to build and deploy configurations.
1.14. How you can use Ruby on Rails
As a developer, you can use Ruby on Rails to:
Write your application:
- Set up a database.
- Create a welcome page.
- Configure your application for OpenShift Container Platform.
- Store your application in Git.
Deploy your application in OpenShift Container Platform:
- Create the database service.
- Create the frontend service.
- Create a route for your application.
Chapter 2. Configuring the Cluster Samples Operator
The Cluster Samples Operator, which operates in the openshift
namespace, installs and updates the Red Hat Enterprise Linux (RHEL)-based OpenShift Container Platform image streams and OpenShift Container Platform templates.
Starting from OpenShift Container Platform 4.13, Cluster Samples Operator is downsized. Cluster Samples Operator will stop providing the following updates for non-Source-to-Image (Non-S2I) image streams and templates:
- new image streams and templates
- updates to the existing image streams and templates unless it is a CVE update
- Cluster Samples Operator will provide support for Non-S2I image streams and templates as per the OpenShift Container Platform lifecycle policy dates and support guidelines.
Cluster Samples Operator will continue to support the S2I builder image streams and templates and accept the updates. S2I image streams and templates include:
- Ruby
- Python
- Node.js
- Perl
- PHP
- HTTPD
- Nginx
- EAP
- Java
- Webserver
- .NET
- Go
- Starting from OpenShift Container Platform 4.16, Cluster Samples Operator will stop managing non-S2I image streams and templates. You can contact the image stream or template owner for any requirements and future plans. In addition, refer to the list of the repositories hosting the image stream or templates.
2.1. Understanding the Cluster Samples Operator
During installation, the Operator creates the default configuration object for itself and then creates the sample image streams and templates, including quick start templates.
To facilitate image stream imports from other registries that require credentials, a cluster administrator can create any additional secrets that contain the content of a Docker config.json
file in the openshift
namespace needed for image import.
The Cluster Samples Operator configuration is a cluster-wide resource, and the deployment is contained within the openshift-cluster-samples-operator
namespace.
The image for the Cluster Samples Operator contains image stream and template definitions for the associated OpenShift Container Platform release. When each sample is created or updated, the Cluster Samples Operator includes an annotation that denotes the version of OpenShift Container Platform. The Operator uses this annotation to ensure that each sample matches the release version. Samples outside of its inventory are ignored, as are skipped samples. Modifications to any samples that are managed by the Operator, where that version annotation is modified or deleted, are reverted automatically.
The Jenkins images are part of the image payload from installation and are tagged into the image streams directly.
The Cluster Samples Operator configuration resource includes a finalizer which cleans up the following upon deletion:
- Operator managed image streams.
- Operator managed templates.
- Operator generated configuration resources.
- Cluster status resources.
Upon deletion of the samples resource, the Cluster Samples Operator recreates the resource using the default configuration.
2.1.1. Cluster Samples Operator’s use of management state
The Cluster Samples Operator is bootstrapped as Managed
by default or if global proxy is configured. In the Managed
state, the Cluster Samples Operator is actively managing its resources and keeping the component active in order to pull sample image streams and images from the registry and ensure that the requisite sample templates are installed.
Certain circumstances result in the Cluster Samples Operator bootstrapping itself as Removed
including:
- If the Cluster Samples Operator cannot reach registry.redhat.io after three minutes on initial startup after a clean installation.
- If the Cluster Samples Operator detects it is on an IPv6 network.
-
If the image controller configuration parameters prevent the creation of image streams by using the default image registry, or by using the image registry specified by the
samplesRegistry
setting.
For OpenShift Container Platform, the default image registry is registry.redhat.io
.
However, if the Cluster Samples Operator detects that it is on an IPv6 network and an OpenShift Container Platform global proxy is configured, then IPv6 check supersedes all the checks. As a result, the Cluster Samples Operator bootstraps itself as Removed
.
IPv6 installations are not currently supported by registry.redhat.io. The Cluster Samples Operator pulls most of the sample image streams and images from registry.redhat.io.
2.1.1.1. Restricted network installation
Boostrapping as Removed
when unable to access registry.redhat.io
facilitates restricted network installations when the network restriction is already in place. Bootstrapping as Removed
when network access is restricted allows the cluster administrator more time to decide if samples are desired, because the Cluster Samples Operator does not submit alerts that sample image stream imports are failing when the management state is set to Removed
. When the Cluster Samples Operator comes up as Managed
and attempts to install sample image streams, it starts alerting two hours after initial installation if there are failing imports.
2.1.1.2. Restricted network installation with initial network access
Conversely, if a cluster that is intended to be a restricted network or disconnected cluster is first installed while network access exists, the Cluster Samples Operator installs the content from registry.redhat.io
since it can access it. If you want the Cluster Samples Operator to still bootstrap as Removed
in order to defer samples installation until you have decided which samples are desired, set up image mirrors, and so on, then follow the instructions for using the Samples Operator with an alternate registry and customizing nodes, both linked in the additional resources section, to override the Cluster Samples Operator default configuration and initially come up as Removed
.
You must put the following additional YAML file in the openshift
directory created by openshift-install create manifest
:
Example Cluster Samples Operator YAML file with managementState: Removed
apiVersion: samples.operator.openshift.io/v1 kind: Config metadata: name: cluster spec: architectures: - x86_64 managementState: Removed
2.1.2. Cluster Samples Operator’s tracking and error recovery of image stream imports
After creation or update of a samples image stream, the Cluster Samples Operator monitors the progress of each image stream tag’s image import.
If an import fails, the Cluster Samples Operator retries the import through the image stream image import API, which is the same API used by the oc import-image
command, approximately every 15 minutes until it sees the import succeed, or if the Cluster Samples Operator’s configuration is changed such that either the image stream is added to the skippedImagestreams
list, or the management state is changed to Removed
.
Additional resources
-
If the Cluster Samples Operator is removed during installation, you can use the Cluster Samples Operator with an alternate registry so content can be imported, and then set the Cluster Samples Operator to
Managed
to get the samples. To ensure the Cluster Samples Operator bootstraps as
Removed
in a restricted network installation with initial network access to defer samples installation until you have decided which samples are desired, follow the instructions for customizing nodes to override the Cluster Samples Operator default configuration and initially come up asRemoved
.- To host samples in your disconnected environment, follow the instructions for using the Cluster Samples Operator with an alternate registry.
2.1.3. Cluster Samples Operator assistance for mirroring
During installation, OpenShift Container Platform creates a config map named imagestreamtag-to-image
in the openshift-cluster-samples-operator
namespace. The imagestreamtag-to-image
config map contains an entry, the populating image, for each image stream tag.
The format of the key for each entry in the data field in the config map is <image_stream_name>_<image_stream_tag_name>
.
During a disconnected installation of OpenShift Container Platform, the status of the Cluster Samples Operator is set to Removed
. If you choose to change it to Managed
, it installs samples.
The use of samples in a network-restricted or discontinued environment may require access to services external to your network. Some example services include: Github, Maven Central, npm, RubyGems, PyPi and others. There might be additional steps to take that allow the cluster samples operators’s objects to reach the services they require.
You can use this config map as a reference for which images need to be mirrored for your image streams to import.
-
While the Cluster Samples Operator is set to
Removed
, you can create your mirrored registry, or determine which existing mirrored registry you want to use. - Mirror the samples you want to the mirrored registry using the new config map as your guide.
-
Add any of the image streams you did not mirror to the
skippedImagestreams
list of the Cluster Samples Operator configuration object. -
Set
samplesRegistry
of the Cluster Samples Operator configuration object to the mirrored registry. -
Then set the Cluster Samples Operator to
Managed
to install the image streams you have mirrored.
See Using Cluster Samples Operator image streams with alternate or mirrored registries for a detailed procedure.
2.2. Cluster Samples Operator configuration parameters
The samples resource offers the following configuration fields:
Parameter | Description |
---|---|
|
|
|
Allows you to specify which registry is accessed by image streams for their image content. Note
Creation or update of RHEL content does not commence if the secret for pull access is not in place when either
Creation or update of RHEL content is not gated by the existence of the pull secret if the |
| Placeholder to choose an architecture type. |
|
Image streams that are in the Cluster Samples Operator’s inventory but that the cluster administrator wants the Operator to ignore or not manage. You can add a list of image stream names to this parameter. For example, |
| Templates that are in the Cluster Samples Operator’s inventory, but that the cluster administrator wants the Operator to ignore or not manage. |
Secret, image stream, and template watch events can come in before the initial samples resource object is created, the Cluster Samples Operator detects and re-queues the event.
2.2.1. Configuration restrictions
When the Cluster Samples Operator starts supporting multiple architectures, the architecture list is not allowed to be changed while in the Managed
state.
To change the architectures values, a cluster administrator must:
-
Mark the
Management State
asRemoved
, saving the change. -
In a subsequent change, edit the architecture and change the
Management State
back toManaged
.
The Cluster Samples Operator still processes secrets while in Removed
state. You can create the secret before switching to Removed
, while in Removed
before switching to Managed
, or after switching to Managed
state. There are delays in creating the samples until the secret event is processed if you create the secret after switching to Managed
. This helps facilitate the changing of the registry, where you choose to remove all the samples before switching to insure a clean slate. Removing all samples before switching is not required.
2.2.2. Conditions
The samples resource maintains the following conditions in its status:
Condition | Description |
---|---|
|
Indicates the samples are created in the |
|
This condition is deprecated in OpenShift Container Platform. |
|
|
|
Indicator that there is a |
| Indicator of which image streams had errors during the image import phase for one of their tags.
|
|
This condition is deprecated in OpenShift Container Platform. |
2.3. Accessing the Cluster Samples Operator configuration
You can configure the Cluster Samples Operator by editing the file with the provided parameters.
Prerequisites
-
Install the OpenShift CLI (
oc
).
Procedure
Access the Cluster Samples Operator configuration:
$ oc edit configs.samples.operator.openshift.io/cluster -o yaml
The Cluster Samples Operator configuration resembles the following example:
apiVersion: samples.operator.openshift.io/v1 kind: Config # ...
2.4. Removing deprecated image stream tags from the Cluster Samples Operator
The Cluster Samples Operator leaves deprecated image stream tags in an image stream because users can have deployments that use the deprecated image stream tags.
You can remove deprecated image stream tags by editing the image stream with the oc tag
command.
Deprecated image stream tags that the samples providers have removed from their image streams are not included on initial installations.
Prerequisites
-
You installed the
oc
CLI.
Procedure
Remove deprecated image stream tags by editing the image stream with the
oc tag
command.$ oc tag -d <image_stream_name:tag>
Example output
Deleted tag default/<image_stream_name:tag>.
Additional resources
- For more information about configuring credentials, see Using image pull secrets.
Chapter 3. Using the Cluster Samples Operator with an alternate registry
You can use the Cluster Samples Operator with an alternate registry by first creating a mirror registry.
You must have access to the internet to obtain the necessary container images. In this procedure, you place the mirror registry on a mirror host that has access to both your network and the internet.
3.1. About the mirror registry
You can mirror the images that are required for OpenShift Container Platform installation and subsequent product updates to a container mirror registry such as Red Hat Quay, JFrog Artifactory, Sonatype Nexus Repository, or Harbor. If you do not have access to a large-scale container registry, you can use the mirror registry for Red Hat OpenShift, a small-scale container registry included with OpenShift Container Platform subscriptions.
You can use any container registry that supports Docker v2-2, such as Red Hat Quay, the mirror registry for Red Hat OpenShift, Artifactory, Sonatype Nexus Repository, or Harbor. Regardless of your chosen registry, the procedure to mirror content from Red Hat hosted sites on the internet to an isolated image registry is the same. After you mirror the content, you configure each cluster to retrieve this content from your mirror registry.
The OpenShift image registry cannot be used as the target registry because it does not support pushing without a tag, which is required during the mirroring process.
If choosing a container registry that is not the mirror registry for Red Hat OpenShift, it must be reachable by every machine in the clusters that you provision. If the registry is unreachable, installation, updating, or normal operations such as workload relocation might fail. For that reason, you must run mirror registries in a highly available way, and the mirror registries must at least match the production availability of your OpenShift Container Platform clusters.
When you populate your mirror registry with OpenShift Container Platform images, you can follow two scenarios. If you have a host that can access both the internet and your mirror registry, but not your cluster nodes, you can directly mirror the content from that machine. This process is referred to as connected mirroring. If you have no such host, you must mirror the images to a file system and then bring that host or removable media into your restricted environment. This process is referred to as disconnected mirroring.
For mirrored registries, to view the source of pulled images, you must review the Trying to access
log entry in the CRI-O logs. Other methods to view the image pull source, such as using the crictl images
command on a node, show the non-mirrored image name, even though the image is pulled from the mirrored location.
Red Hat does not test third party registries with OpenShift Container Platform.
Additional information
For information on viewing the CRI-O logs to view the image source, see Viewing the image pull source.
3.1.1. Preparing the mirror host
Before you create the mirror registry, you must prepare the mirror host.
3.1.2. Installing the OpenShift CLI by downloading the binary
You can install the OpenShift CLI (oc
) to interact with OpenShift Container Platform from a command-line interface. You can install oc
on Linux, Windows, or macOS.
If you installed an earlier version of oc
, you cannot use it to complete all of the commands in OpenShift Container Platform 4.15. Download and install the new version of oc
.
Installing the OpenShift CLI on Linux
You can install the OpenShift CLI (oc
) binary on Linux by using the following procedure.
Procedure
- Navigate to the OpenShift Container Platform downloads page on the Red Hat Customer Portal.
- Select the architecture from the Product Variant drop-down list.
- Select the appropriate version from the Version drop-down list.
- Click Download Now next to the OpenShift v4.15 Linux Client entry and save the file.
Unpack the archive:
$ tar xvf <file>
Place the
oc
binary in a directory that is on yourPATH
.To check your
PATH
, execute the following command:$ echo $PATH
Verification
After you install the OpenShift CLI, it is available using the
oc
command:$ oc <command>
Installing the OpenShift CLI on Windows
You can install the OpenShift CLI (oc
) binary on Windows by using the following procedure.
Procedure
- Navigate to the OpenShift Container Platform downloads page on the Red Hat Customer Portal.
- Select the appropriate version from the Version drop-down list.
- Click Download Now next to the OpenShift v4.15 Windows Client entry and save the file.
- Unzip the archive with a ZIP program.
Move the
oc
binary to a directory that is on yourPATH
.To check your
PATH
, open the command prompt and execute the following command:C:\> path
Verification
After you install the OpenShift CLI, it is available using the
oc
command:C:\> oc <command>
Installing the OpenShift CLI on macOS
You can install the OpenShift CLI (oc
) binary on macOS by using the following procedure.
Procedure
- Navigate to the OpenShift Container Platform downloads page on the Red Hat Customer Portal.
- Select the appropriate version from the Version drop-down list.
Click Download Now next to the OpenShift v4.15 macOS Client entry and save the file.
NoteFor macOS arm64, choose the OpenShift v4.15 macOS arm64 Client entry.
- Unpack and unzip the archive.
Move the
oc
binary to a directory on your PATH.To check your
PATH
, open a terminal and execute the following command:$ echo $PATH
Verification
Verify your installation by using an
oc
command:$ oc <command>
3.2. Configuring credentials that allow images to be mirrored
Create a container image registry credentials file that allows mirroring images from Red Hat to your mirror.
Prerequisites
- You configured a mirror registry to use in your disconnected environment.
Procedure
Complete the following steps on the installation host:
-
Download your
registry.redhat.io
pull secret from Red Hat OpenShift Cluster Manager. Make a copy of your pull secret in JSON format:
$ cat ./pull-secret | jq . > <path>/<pull_secret_file_in_json> 1
- 1
- Specify the path to the folder to store the pull secret in and a name for the JSON file that you create.
The contents of the file resemble the following example:
{ "auths": { "cloud.openshift.com": { "auth": "b3BlbnNo...", "email": "you@example.com" }, "quay.io": { "auth": "b3BlbnNo...", "email": "you@example.com" }, "registry.connect.redhat.com": { "auth": "NTE3Njg5Nj...", "email": "you@example.com" }, "registry.redhat.io": { "auth": "NTE3Njg5Nj...", "email": "you@example.com" } } }
Generate the base64-encoded user name and password or token for your mirror registry:
$ echo -n '<user_name>:<password>' | base64 -w0 1 BGVtbYk3ZHAtqXs=
- 1
- For
<user_name>
and<password>
, specify the user name and password that you configured for your registry.
Edit the JSON file and add a section that describes your registry to it:
"auths": { "<mirror_registry>": { 1 "auth": "<credentials>", 2 "email": "you@example.com" } },
The file resembles the following example:
{ "auths": { "registry.example.com": { "auth": "BGVtbYk3ZHAtqXs=", "email": "you@example.com" }, "cloud.openshift.com": { "auth": "b3BlbnNo...", "email": "you@example.com" }, "quay.io": { "auth": "b3BlbnNo...", "email": "you@example.com" }, "registry.connect.redhat.com": { "auth": "NTE3Njg5Nj...", "email": "you@example.com" }, "registry.redhat.io": { "auth": "NTE3Njg5Nj...", "email": "you@example.com" } } }
3.3. Mirroring the OpenShift Container Platform image repository
Mirror the OpenShift Container Platform image repository to your registry to use during cluster installation or upgrade.
Prerequisites
- Your mirror host has access to the internet.
- You configured a mirror registry to use in your restricted network and can access the certificate and credentials that you configured.
- You downloaded the pull secret from Red Hat OpenShift Cluster Manager and modified it to include authentication to your mirror repository.
- If you use self-signed certificates, you have specified a Subject Alternative Name in the certificates.
Procedure
Complete the following steps on the mirror host:
- Review the OpenShift Container Platform downloads page to determine the version of OpenShift Container Platform that you want to install and determine the corresponding tag on the Repository Tags page.
Set the required environment variables:
Export the release version:
$ OCP_RELEASE=<release_version>
For
<release_version>
, specify the tag that corresponds to the version of OpenShift Container Platform to install, such as4.5.4
.Export the local registry name and host port:
$ LOCAL_REGISTRY='<local_registry_host_name>:<local_registry_host_port>'
For
<local_registry_host_name>
, specify the registry domain name for your mirror repository, and for<local_registry_host_port>
, specify the port that it serves content on.Export the local repository name:
$ LOCAL_REPOSITORY='<local_repository_name>'
For
<local_repository_name>
, specify the name of the repository to create in your registry, such asocp4/openshift4
.Export the name of the repository to mirror:
$ PRODUCT_REPO='openshift-release-dev'
For a production release, you must specify
openshift-release-dev
.Export the path to your registry pull secret:
$ LOCAL_SECRET_JSON='<path_to_pull_secret>'
For
<path_to_pull_secret>
, specify the absolute path to and file name of the pull secret for your mirror registry that you created.Export the release mirror:
$ RELEASE_NAME="ocp-release"
For a production release, you must specify
ocp-release
.Export the type of architecture for your cluster:
$ ARCHITECTURE=<cluster_architecture> 1
- 1
- Specify the architecture of the cluster, such as
x86_64
,aarch64
,s390x
, orppc64le
.
Export the path to the directory to host the mirrored images:
$ REMOVABLE_MEDIA_PATH=<path> 1
- 1
- Specify the full path, including the initial forward slash (/) character.
Mirror the version images to the mirror registry:
If your mirror host does not have internet access, take the following actions:
- Connect the removable media to a system that is connected to the internet.
Review the images and configuration manifests to mirror:
$ oc adm release mirror -a ${LOCAL_SECRET_JSON} \ --from=quay.io/${PRODUCT_REPO}/${RELEASE_NAME}:${OCP_RELEASE}-${ARCHITECTURE} \ --to=${LOCAL_REGISTRY}/${LOCAL_REPOSITORY} \ --to-release-image=${LOCAL_REGISTRY}/${LOCAL_REPOSITORY}:${OCP_RELEASE}-${ARCHITECTURE} --dry-run
-
Record the entire
imageContentSources
section from the output of the previous command. The information about your mirrors is unique to your mirrored repository, and you must add theimageContentSources
section to theinstall-config.yaml
file during installation. Mirror the images to a directory on the removable media:
$ oc adm release mirror -a ${LOCAL_SECRET_JSON} --to-dir=${REMOVABLE_MEDIA_PATH}/mirror quay.io/${PRODUCT_REPO}/${RELEASE_NAME}:${OCP_RELEASE}-${ARCHITECTURE}
Take the media to the restricted network environment and upload the images to the local container registry.
$ oc image mirror -a ${LOCAL_SECRET_JSON} --from-dir=${REMOVABLE_MEDIA_PATH}/mirror "file://openshift/release:${OCP_RELEASE}*" ${LOCAL_REGISTRY}/${LOCAL_REPOSITORY} 1
- 1
- For
REMOVABLE_MEDIA_PATH
, you must use the same path that you specified when you mirrored the images.
ImportantRunning
oc image mirror
might result in the following error:error: unable to retrieve source image
. This error occurs when image indexes include references to images that no longer exist on the image registry. Image indexes might retain older references to allow users running those images an upgrade path to newer points on the upgrade graph. As a temporary workaround, you can use the--skip-missing
option to bypass the error and continue downloading the image index. For more information, see Service Mesh Operator mirroring failed.
If the local container registry is connected to the mirror host, take the following actions:
Directly push the release images to the local registry by using following command:
$ oc adm release mirror -a ${LOCAL_SECRET_JSON} \ --from=quay.io/${PRODUCT_REPO}/${RELEASE_NAME}:${OCP_RELEASE}-${ARCHITECTURE} \ --to=${LOCAL_REGISTRY}/${LOCAL_REPOSITORY} \ --to-release-image=${LOCAL_REGISTRY}/${LOCAL_REPOSITORY}:${OCP_RELEASE}-${ARCHITECTURE}
This command pulls the release information as a digest, and its output includes the
imageContentSources
data that you require when you install your cluster.Record the entire
imageContentSources
section from the output of the previous command. The information about your mirrors is unique to your mirrored repository, and you must add theimageContentSources
section to theinstall-config.yaml
file during installation.NoteThe image name gets patched to Quay.io during the mirroring process, and the podman images will show Quay.io in the registry on the bootstrap virtual machine.
To create the installation program that is based on the content that you mirrored, extract it and pin it to the release:
If your mirror host does not have internet access, run the following command:
$ oc adm release extract -a ${LOCAL_SECRET_JSON} --icsp-file=<file> --command=openshift-install "${LOCAL_REGISTRY}/${LOCAL_REPOSITORY}:${OCP_RELEASE}-${ARCHITECTURE}" \ --insecure=true 1
- 1
- Optional: If you do not want to configure trust for the target registry, add the
--insecure=true
flag.
If the local container registry is connected to the mirror host, run the following command:
$ oc adm release extract -a ${LOCAL_SECRET_JSON} --command=openshift-install "${LOCAL_REGISTRY}/${LOCAL_REPOSITORY}:${OCP_RELEASE}-${ARCHITECTURE}"
ImportantTo ensure that you use the correct images for the version of OpenShift Container Platform that you selected, you must extract the installation program from the mirrored content.
You must perform this step on a machine with an active internet connection.
For clusters using installer-provisioned infrastructure, run the following command:
$ openshift-install
3.4. Using Cluster Samples Operator image streams with alternate or mirrored registries
Most image streams in the openshift
namespace managed by the Cluster Samples Operator point to images located in the Red Hat registry at registry.redhat.io.
The cli
, installer
, must-gather
, and tests
image streams, while part of the install payload, are not managed by the Cluster Samples Operator. These are not addressed in this procedure.
The Cluster Samples Operator must be set to Managed
in a disconnected environment. To install the image streams, you have a mirrored registry.
Prerequisites
-
Access to the cluster as a user with the
cluster-admin
role. - Create a pull secret for your mirror registry.
Procedure
Access the images of a specific image stream to mirror, for example:
$ oc get is <imagestream> -n openshift -o json | jq .spec.tags[].from.name | grep registry.redhat.io
Mirror images from registry.redhat.io associated with any image streams you need
$ oc image mirror registry.redhat.io/rhscl/ruby-25-rhel7:latest ${MIRROR_ADDR}/rhscl/ruby-25-rhel7:latest
Create the cluster’s image configuration object:
$ oc create configmap registry-config --from-file=${MIRROR_ADDR_HOSTNAME}..5000=$path/ca.crt -n openshift-config
Add the required trusted CAs for the mirror in the cluster’s image configuration object:
$ oc patch image.config.openshift.io/cluster --patch '{"spec":{"additionalTrustedCA":{"name":"registry-config"}}}' --type=merge
Update the
samplesRegistry
field in the Cluster Samples Operator configuration object to contain thehostname
portion of the mirror location defined in the mirror configuration:$ oc edit configs.samples.operator.openshift.io -n openshift-cluster-samples-operator
NoteThis is required because the image stream import process does not use the mirror or search mechanism at this time.
Add any image streams that are not mirrored into the
skippedImagestreams
field of the Cluster Samples Operator configuration object. Or if you do not want to support any of the sample image streams, set the Cluster Samples Operator toRemoved
in the Cluster Samples Operator configuration object.NoteThe Cluster Samples Operator issues alerts if image stream imports are failing but the Cluster Samples Operator is either periodically retrying or does not appear to be retrying them.
Many of the templates in the
openshift
namespace reference the image streams. So usingRemoved
to purge both the image streams and templates will eliminate the possibility of attempts to use them if they are not functional because of any missing image streams.
3.4.1. Cluster Samples Operator assistance for mirroring
During installation, OpenShift Container Platform creates a config map named imagestreamtag-to-image
in the openshift-cluster-samples-operator
namespace. The imagestreamtag-to-image
config map contains an entry, the populating image, for each image stream tag.
The format of the key for each entry in the data field in the config map is <image_stream_name>_<image_stream_tag_name>
.
During a disconnected installation of OpenShift Container Platform, the status of the Cluster Samples Operator is set to Removed
. If you choose to change it to Managed
, it installs samples.
The use of samples in a network-restricted or discontinued environment may require access to services external to your network. Some example services include: Github, Maven Central, npm, RubyGems, PyPi and others. There might be additional steps to take that allow the cluster samples operators’s objects to reach the services they require.
You can use this config map as a reference for which images need to be mirrored for your image streams to import.
-
While the Cluster Samples Operator is set to
Removed
, you can create your mirrored registry, or determine which existing mirrored registry you want to use. - Mirror the samples you want to the mirrored registry using the new config map as your guide.
-
Add any of the image streams you did not mirror to the
skippedImagestreams
list of the Cluster Samples Operator configuration object. -
Set
samplesRegistry
of the Cluster Samples Operator configuration object to the mirrored registry. -
Then set the Cluster Samples Operator to
Managed
to install the image streams you have mirrored.
See Using Cluster Samples Operator image streams with alternate or mirrored registries for a detailed procedure.
Chapter 4. Creating images
Learn how to create your own container images, based on pre-built images that are ready to help you. The process includes learning best practices for writing images, defining metadata for images, testing images, and using a custom builder workflow to create images to use with OpenShift Container Platform. After you create an image, you can push it to the OpenShift image registry.
4.1. Learning container best practices
When creating container images to run on OpenShift Container Platform there are a number of best practices to consider as an image author to ensure a good experience for consumers of those images. Because images are intended to be immutable and used as-is, the following guidelines help ensure that your images are highly consumable and easy to use on OpenShift Container Platform.
4.1.1. General container image guidelines
The following guidelines apply when creating a container image in general, and are independent of whether the images are used on OpenShift Container Platform.
Reuse images
Wherever possible, base your image on an appropriate upstream image using the FROM
statement. This ensures your image can easily pick up security fixes from an upstream image when it is updated, rather than you having to update your dependencies directly.
In addition, use tags in the FROM
instruction, for example, rhel:rhel7
, to make it clear to users exactly which version of an image your image is based on. Using a tag other than latest
ensures your image is not subjected to breaking changes that might go into the latest
version of an upstream image.
Maintain compatibility within tags
When tagging your own images, try to maintain backwards compatibility within a tag. For example, if you provide an image named image
and it currently includes version 1.0
, you might provide a tag of image:v1
. When you update the image, as long as it continues to be compatible with the original image, you can continue to tag the new image image:v1
, and downstream consumers of this tag are able to get updates without being broken.
If you later release an incompatible update, then switch to a new tag, for example image:v2
. This allows downstream consumers to move up to the new version at will, but not be inadvertently broken by the new incompatible image. Any downstream consumer using image:latest
takes on the risk of any incompatible changes being introduced.
Avoid multiple processes
Do not start multiple services, such as a database and SSHD
, inside one container. This is not necessary because containers are lightweight and can be easily linked together for orchestrating multiple processes. OpenShift Container Platform allows you to easily colocate and co-manage related images by grouping them into a single pod.
This colocation ensures the containers share a network namespace and storage for communication. Updates are also less disruptive as each image can be updated less frequently and independently. Signal handling flows are also clearer with a single process as you do not have to manage routing signals to spawned processes.
Use exec
in wrapper scripts
Many images use wrapper scripts to do some setup before starting a process for the software being run. If your image uses such a script, that script uses exec
so that the script’s process is replaced by your software. If you do not use exec
, then signals sent by your container runtime go to your wrapper script instead of your software’s process. This is not what you want.
If you have a wrapper script that starts a process for some server. You start your container, for example, using podman run -i
, which runs the wrapper script, which in turn starts your process. If you want to close your container with CTRL+C
. If your wrapper script used exec
to start the server process, podman
sends SIGINT to the server process, and everything works as you expect. If you did not use exec
in your wrapper script, podman
sends SIGINT to the process for the wrapper script and your process keeps running like nothing happened.
Also note that your process runs as PID 1
when running in a container. This means that if your main process terminates, the entire container is stopped, canceling any child processes you launched from your PID 1
process.
Clean temporary files
Remove all temporary files you create during the build process. This also includes any files added with the ADD
command. For example, run the yum clean
command after performing yum install
operations.
You can prevent the yum
cache from ending up in an image layer by creating your RUN
statement as follows:
RUN yum -y install mypackage && yum -y install myotherpackage && yum clean all -y
Note that if you instead write:
RUN yum -y install mypackage RUN yum -y install myotherpackage && yum clean all -y
Then the first yum
invocation leaves extra files in that layer, and these files cannot be removed when the yum clean
operation is run later. The extra files are not visible in the final image, but they are present in the underlying layers.
The current container build process does not allow a command run in a later layer to shrink the space used by the image when something was removed in an earlier layer. However, this may change in the future. This means that if you perform an rm
command in a later layer, although the files are hidden it does not reduce the overall size of the image to be downloaded. Therefore, as with the yum clean
example, it is best to remove files in the same command that created them, where possible, so they do not end up written to a layer.
In addition, performing multiple commands in a single RUN
statement reduces the number of layers in your image, which improves download and extraction time.
Place instructions in the proper order
The container builder reads the Dockerfile
and runs the instructions from top to bottom. Every instruction that is successfully executed creates a layer which can be reused the next time this or another image is built. It is very important to place instructions that rarely change at the top of your Dockerfile
. Doing so ensures the next builds of the same image are very fast because the cache is not invalidated by upper layer changes.
For example, if you are working on a Dockerfile
that contains an ADD
command to install a file you are iterating on, and a RUN
command to yum install
a package, it is best to put the ADD
command last:
FROM foo RUN yum -y install mypackage && yum clean all -y ADD myfile /test/myfile
This way each time you edit myfile
and rerun podman build
or docker build
, the system reuses the cached layer for the yum
command and only generates the new layer for the ADD
operation.
If instead you wrote the Dockerfile
as:
FROM foo ADD myfile /test/myfile RUN yum -y install mypackage && yum clean all -y
Then each time you changed myfile
and reran podman build
or docker build
, the ADD
operation would invalidate the RUN
layer cache, so the yum
operation must be rerun as well.
Mark important ports
The EXPOSE instruction makes a port in the container available to the host system and other containers. While it is possible to specify that a port should be exposed with a podman run
invocation, using the EXPOSE instruction in a Dockerfile
makes it easier for both humans and software to use your image by explicitly declaring the ports your software needs to run:
-
Exposed ports show up under
podman ps
associated with containers created from your image. -
Exposed ports are present in the metadata for your image returned by
podman inspect
. - Exposed ports are linked when you link one container to another.
Set environment variables
It is good practice to set environment variables with the ENV
instruction. One example is to set the version of your project. This makes it easy for people to find the version without looking at the Dockerfile
. Another example is advertising a path on the system that could be used by another process, such as JAVA_HOME
.
Avoid default passwords
Avoid setting default passwords. Many people extend the image and forget to remove or change the default password. This can lead to security issues if a user in production is assigned a well-known password. Passwords are configurable using an environment variable instead.
If you do choose to set a default password, ensure that an appropriate warning message is displayed when the container is started. The message should inform the user of the value of the default password and explain how to change it, such as what environment variable to set.
Avoid sshd
It is best to avoid running sshd
in your image. You can use the podman exec
or docker exec
command to access containers that are running on the local host. Alternatively, you can use the oc exec
command or the oc rsh
command to access containers that are running on the OpenShift Container Platform cluster. Installing and running sshd
in your image opens up additional vectors for attack and requirements for security patching.
Use volumes for persistent data
Images use a volume for persistent data. This way OpenShift Container Platform mounts the network storage to the node running the container, and if the container moves to a new node the storage is reattached to that node. By using the volume for all persistent storage needs, the content is preserved even if the container is restarted or moved. If your image writes data to arbitrary locations within the container, that content could not be preserved.
All data that needs to be preserved even after the container is destroyed must be written to a volume. Container engines support a readonly
flag for containers, which can be used to strictly enforce good practices about not writing data to ephemeral storage in a container. Designing your image around that capability now makes it easier to take advantage of it later.
Explicitly defining volumes in your Dockerfile
makes it easy for consumers of the image to understand what volumes they must define when running your image.
See the Kubernetes documentation for more information on how volumes are used in OpenShift Container Platform.
Even with persistent volumes, each instance of your image has its own volume, and the filesystem is not shared between instances. This means the volume cannot be used to share state in a cluster.
4.1.2. OpenShift Container Platform-specific guidelines
The following are guidelines that apply when creating container images specifically for use on OpenShift Container Platform.
4.1.2.1. Enable images for source-to-image (S2I)
For images that are intended to run application code provided by a third party, such as a Ruby image designed to run Ruby code provided by a developer, you can enable your image to work with the Source-to-Image (S2I) build tool. S2I is a framework that makes it easy to write images that take application source code as an input and produce a new image that runs the assembled application as output.
4.1.2.2. Support arbitrary user ids
By default, OpenShift Container Platform runs containers using an arbitrarily assigned user ID. This provides additional security against processes escaping the container due to a container engine vulnerability and thereby achieving escalated permissions on the host node.
For an image to support running as an arbitrary user, directories and files that are written to by processes in the image must be owned by the root group and be read/writable by that group. Files to be executed must also have group execute permissions.
Adding the following to your Dockerfile sets the directory and file permissions to allow users in the root group to access them in the built image:
RUN chgrp -R 0 /some/directory && \ chmod -R g=u /some/directory
Because the container user is always a member of the root group, the container user can read and write these files.
Care must be taken when altering the directories and file permissions of the sensitive areas of a container. If applied to sensitive areas, such as the /etc/passwd
file, such changes can allow the modification of these files by unintended users, potentially exposing the container or host. CRI-O supports the insertion of arbitrary user IDs into a container’s /etc/passwd
file. As such, changing permissions is never required.
Additionally, the /etc/passwd
file should not exist in any container image. If it does, the CRI-O container runtime will fail to inject a random UID into the /etc/passwd
file. In such cases, the container might face challenges in resolving the active UID. Failing to meet this requirement could impact the functionality of certain containerized applications.
In addition, the processes running in the container must not listen on privileged ports, ports below 1024, since they are not running as a privileged user.
If your S2I image does not include a USER
declaration with a numeric user, your builds fail by default. To allow images that use either named users or the root 0
user to build in OpenShift Container Platform, you can add the project’s builder service account, system:serviceaccount:<your-project>:builder
, to the anyuid
security context constraint (SCC). Alternatively, you can allow all images to run as any user.
4.1.2.3. Use services for inter-image communication
For cases where your image needs to communicate with a service provided by another image, such as a web front end image that needs to access a database image to store and retrieve data, your image consumes an OpenShift Container Platform service. Services provide a static endpoint for access which does not change as containers are stopped, started, or moved. In addition, services provide load balancing for requests.
4.1.2.4. Provide common libraries
For images that are intended to run application code provided by a third party, ensure that your image contains commonly used libraries for your platform. In particular, provide database drivers for common databases used with your platform. For example, provide JDBC drivers for MySQL and PostgreSQL if you are creating a Java framework image. Doing so prevents the need for common dependencies to be downloaded during application assembly time, speeding up application image builds. It also simplifies the work required by application developers to ensure all of their dependencies are met.
4.1.2.5. Use environment variables for configuration
Users of your image are able to configure it without having to create a downstream image based on your image. This means that the runtime configuration is handled using environment variables. For a simple configuration, the running process can consume the environment variables directly. For a more complicated configuration or for runtimes which do not support this, configure the runtime by defining a template configuration file that is processed during startup. During this processing, values supplied using environment variables can be substituted into the configuration file or used to make decisions about what options to set in the configuration file.
It is also possible and recommended to pass secrets such as certificates and keys into the container using environment variables. This ensures that the secret values do not end up committed in an image and leaked into a container image registry.
Providing environment variables allows consumers of your image to customize behavior, such as database settings, passwords, and performance tuning, without having to introduce a new layer on top of your image. Instead, they can simply define environment variable values when defining a pod and change those settings without rebuilding the image.
For extremely complex scenarios, configuration can also be supplied using volumes that would be mounted into the container at runtime. However, if you elect to do it this way you must ensure that your image provides clear error messages on startup when the necessary volume or configuration is not present.
This topic is related to the Using Services for Inter-image Communication topic in that configuration like datasources are defined in terms of environment variables that provide the service endpoint information. This allows an application to dynamically consume a datasource service that is defined in the OpenShift Container Platform environment without modifying the application image.
In addition, tuning is done by inspecting the cgroups
settings for the container. This allows the image to tune itself to the available memory, CPU, and other resources. For example, Java-based images tune their heap based on the cgroup
maximum memory parameter to ensure they do not exceed the limits and get an out-of-memory error.
4.1.2.6. Set image metadata
Defining image metadata helps OpenShift Container Platform better consume your container images, allowing OpenShift Container Platform to create a better experience for developers using your image. For example, you can add metadata to provide helpful descriptions of your image, or offer suggestions on other images that are needed.
4.1.2.7. Clustering
You must fully understand what it means to run multiple instances of your image. In the simplest case, the load balancing function of a service handles routing traffic to all instances of your image. However, many frameworks must share information to perform leader election or failover state; for example, in session replication.
Consider how your instances accomplish this communication when running in OpenShift Container Platform. Although pods can communicate directly with each other, their IP addresses change anytime the pod starts, stops, or is moved. Therefore, it is important for your clustering scheme to be dynamic.
4.1.2.8. Logging
It is best to send all logging to standard out. OpenShift Container Platform collects standard out from containers and sends it to the centralized logging service where it can be viewed. If you must separate log content, prefix the output with an appropriate keyword, which makes it possible to filter the messages.
If your image logs to a file, users must use manual operations to enter the running container and retrieve or view the log file.
4.1.2.9. Liveness and readiness probes
Document example liveness and readiness probes that can be used with your image. These probes allow users to deploy your image with confidence that traffic is not be routed to the container until it is prepared to handle it, and that the container is restarted if the process gets into an unhealthy state.
4.1.2.10. Templates
Consider providing an example template with your image. A template gives users an easy way to quickly get your image deployed with a working configuration. Your template must include the liveness and readiness probes you documented with the image, for completeness.
4.2. Including metadata in images
Defining image metadata helps OpenShift Container Platform better consume your container images, allowing OpenShift Container Platform to create a better experience for developers using your image. For example, you can add metadata to provide helpful descriptions of your image, or offer suggestions on other images that may also be needed.
This topic only defines the metadata needed by the current set of use cases. Additional metadata or use cases may be added in the future.
4.2.1. Defining image metadata
You can use the LABEL
instruction in a Dockerfile
to define image metadata. Labels are similar to environment variables in that they are key value pairs attached to an image or a container. Labels are different from environment variable in that they are not visible to the running application and they can also be used for fast look-up of images and containers.
Docker documentation for more information on the LABEL
instruction.
The label names are typically namespaced. The namespace is set accordingly to reflect the project that is going to pick up the labels and use them. For OpenShift Container Platform the namespace is set to io.openshift
and for Kubernetes the namespace is io.k8s
.
See the Docker custom metadata documentation for details about the format.
Variable | Description |
---|---|
| This label contains a list of tags represented as a list of comma-separated string values. The tags are the way to categorize the container images into broad areas of functionality. Tags help UI and generation tools to suggest relevant container images during the application creation process. LABEL io.openshift.tags mongodb,mongodb24,nosql |
|
Specifies a list of tags that the generation tools and the UI uses to provide relevant suggestions if you do not have the container images with specified tags already. For example, if the container image wants LABEL io.openshift.wants mongodb,redis |
| This label can be used to give the container image consumers more detailed information about the service or functionality this image provides. The UI can then use this description together with the container image name to provide more human friendly information to end users. LABEL io.k8s.description The MySQL 5.5 Server with master-slave replication support |
|
An image can use this variable to suggest that it does not support scaling. The UI then communicates this to consumers of that image. Being not-scalable means that the value of LABEL io.openshift.non-scalable true |
| This label suggests how much resources the container image needs to work properly. The UI can warn the user that deploying this container image may exceed their user quota. The values must be compatible with Kubernetes quantity. LABEL io.openshift.min-memory 16Gi LABEL io.openshift.min-cpu 4 |
4.3. Creating images from source code with source-to-image
Source-to-image (S2I) is a framework that makes it easy to write images that take application source code as an input and produce a new image that runs the assembled application as output.
The main advantage of using S2I for building reproducible container images is the ease of use for developers. As a builder image author, you must understand two basic concepts in order for your images to provide the best S2I performance, the build process and S2I scripts.
4.3.1. Understanding the source-to-image build process
The build process consists of the following three fundamental elements, which are combined into a final container image:
- Sources
- Source-to-image (S2I) scripts
- Builder image
S2I generates a Dockerfile with the builder image as the first FROM
instruction. The Dockerfile generated by S2I is then passed to Buildah.
4.3.2. How to write source-to-image scripts
You can write source-to-image (S2I) scripts in any programming language, as long as the scripts are executable inside the builder image. S2I supports multiple options providing assemble
/run
/save-artifacts
scripts. All of these locations are checked on each build in the following order:
- A script specified in the build configuration.
-
A script found in the application source
.s2i/bin
directory. -
A script found at the default image URL with the
io.openshift.s2i.scripts-url
label.
Both the io.openshift.s2i.scripts-url
label specified in the image and the script specified in a build configuration can take one of the following forms:
-
image:///path_to_scripts_dir
: absolute path inside the image to a directory where the S2I scripts are located. -
file:///path_to_scripts_dir
: relative or absolute path to a directory on the host where the S2I scripts are located. -
http(s)://path_to_scripts_dir
: URL to a directory where the S2I scripts are located.
Script | Description |
---|---|
|
The
|
|
The |
|
The
These dependencies are gathered into a |
|
The |
|
The
Note
The suggested location to put the test application built by your |
Example S2I scripts
The following example S2I scripts are written in Bash. Each example assumes its tar
contents are unpacked into the /tmp/s2i
directory.
assemble
script:
#!/bin/bash # restore build artifacts if [ "$(ls /tmp/s2i/artifacts/ 2>/dev/null)" ]; then mv /tmp/s2i/artifacts/* $HOME/. fi # move the application source mv /tmp/s2i/src $HOME/src # build application artifacts pushd ${HOME} make all # install the artifacts make install popd
run
script:
#!/bin/bash # run the application /opt/application/run.sh
save-artifacts
script:
#!/bin/bash pushd ${HOME} if [ -d deps ]; then # all deps contents to tar stream tar cf - deps fi popd
usage
script:
#!/bin/bash # inform the user how to use the image cat <<EOF This is a S2I sample builder image, to use it, install https://github.com/openshift/source-to-image EOF
Additional resources
4.4. About testing source-to-image images
As an Source-to-Image (S2I) builder image author, you can test your S2I image locally and use the OpenShift Container Platform build system for automated testing and continuous integration.
S2I requires the assemble
and run
scripts to be present to successfully run the S2I build. Providing the save-artifacts
script reuses the build artifacts, and providing the usage
script ensures that usage information is printed to console when someone runs the container image outside of the S2I.
The goal of testing an S2I image is to make sure that all of these described commands work properly, even if the base container image has changed or the tooling used by the commands was updated.
4.4.1. Understanding testing requirements
The standard location for the test
script is test/run
. This script is invoked by the OpenShift Container Platform S2I image builder and it could be a simple Bash script or a static Go binary.
The test/run
script performs the S2I build, so you must have the S2I binary available in your $PATH
. If required, follow the installation instructions in the S2I README.
S2I combines the application source code and builder image, so to test it you need a sample application source to verify that the source successfully transforms into a runnable container image. The sample application should be simple, but it should exercise the crucial steps of assemble
and run
scripts.
4.4.2. Generating scripts and tools
The S2I tooling comes with powerful generation tools to speed up the process of creating a new S2I image. The s2i create
command produces all the necessary S2I scripts and testing tools along with the Makefile
:
$ s2i create <image_name> <destination_directory>
The generated test/run
script must be adjusted to be useful, but it provides a good starting point to begin developing.
The test/run
script produced by the s2i create
command requires that the sample application sources are inside the test/test-app
directory.
4.4.3. Testing locally
The easiest way to run the S2I image tests locally is to use the generated Makefile
.
If you did not use the s2i create
command, you can copy the following Makefile
template and replace the IMAGE_NAME
parameter with your image name.
Sample Makefile
IMAGE_NAME = openshift/ruby-20-centos7 CONTAINER_ENGINE := $(shell command -v podman 2> /dev/null | echo docker) build: ${CONTAINER_ENGINE} build -t $(IMAGE_NAME) . .PHONY: test test: ${CONTAINER_ENGINE} build -t $(IMAGE_NAME)-candidate . IMAGE_NAME=$(IMAGE_NAME)-candidate test/run
4.4.4. Basic testing workflow
The test
script assumes you have already built the image you want to test. If required, first build the S2I image. Run one of the following commands:
If you use Podman, run the following command:
$ podman build -t <builder_image_name>
If you use Docker, run the following command:
$ docker build -t <builder_image_name>
The following steps describe the default workflow to test S2I image builders:
Verify the
usage
script is working:If you use Podman, run the following command:
$ podman run <builder_image_name> .
If you use Docker, run the following command:
$ docker run <builder_image_name> .
Build the image:
$ s2i build file:///path-to-sample-app _<BUILDER_IMAGE_NAME>_ _<OUTPUT_APPLICATION_IMAGE_NAME>_
-
Optional: if you support
save-artifacts
, run step 2 once again to verify that saving and restoring artifacts works properly. Run the container:
If you use Podman, run the following command:
$ podman run <output_application_image_name>
If you use Docker, run the following command:
$ docker run <output_application_image_name>
- Verify the container is running and the application is responding.
Running these steps is generally enough to tell if the builder image is working as expected.
4.4.5. Using OpenShift Container Platform for building the image
Once you have a Dockerfile
and the other artifacts that make up your new S2I builder image, you can put them in a git repository and use OpenShift Container Platform to build and push the image. Define a Docker build that points to your repository.
If your OpenShift Container Platform instance is hosted on a public IP address, the build can be triggered each time you push into your S2I builder image GitHub repository.
You can also use the ImageChangeTrigger
to trigger a rebuild of your applications that are based on the S2I builder image you updated.
Chapter 5. Managing images
5.1. Managing images overview
With OpenShift Container Platform you can interact with images and set up image streams, depending on where the registries of the images are located, any authentication requirements around those registries, and how you want your builds and deployments to behave.
5.1.1. Images overview
An image stream comprises any number of container images identified by tags. It presents a single virtual view of related images, similar to a container image repository.
By watching an image stream, builds and deployments can receive notifications when new images are added or modified and react by performing a build or deployment, respectively.
5.2. Tagging images
The following sections provide an overview and instructions for using image tags in the context of container images for working with OpenShift Container Platform image streams and their tags.
5.2.1. Image tags
An image tag is a label applied to a container image in a repository that distinguishes a specific image from other images in an image stream. Typically, the tag represents a version number of some sort. For example, here :v3.11.59-2
is the tag:
registry.access.redhat.com/openshift3/jenkins-2-rhel7:v3.11.59-2
You can add additional tags to an image. For example, an image might be assigned the tags :v3.11.59-2
and :latest
.
OpenShift Container Platform provides the oc tag
command, which is similar to the docker tag
command, but operates on image streams instead of directly on images.
5.2.2. Image tag conventions
Images evolve over time and their tags reflect this. Generally, an image tag always points to the latest image built.
If there is too much information embedded in a tag name, like v2.0.1-may-2019
, the tag points to just one revision of an image and is never updated. Using default image pruning options, such an image is never removed. In very large clusters, the schema of creating new tags for every revised image could eventually fill up the etcd datastore with excess tag metadata for images that are long outdated.
If the tag is named v2.0
, image revisions are more likely. This results in longer tag history and, therefore, the image pruner is more likely to remove old and unused images.
Although tag naming convention is up to you, here are a few examples in the format <image_name>:<image_tag>
:
Description | Example |
---|---|
Revision |
|
Architecture |
|
Base image |
|
Latest (potentially unstable) |
|
Latest stable |
|
If you require dates in tag names, periodically inspect old and unsupported images and istags
and remove them. Otherwise, you can experience increasing resource usage caused by retaining old images.
5.2.3. Adding tags to image streams
An image stream in OpenShift Container Platform comprises zero or more container images identified by tags.
There are different types of tags available. The default behavior uses a permanent
tag, which points to a specific image in time. If the permanent
tag is in use and the source changes, the tag does not change for the destination.
A tracking
tag means the destination tag’s metadata is updated during the import of the source tag.
Procedure
You can add tags to an image stream using the
oc tag
command:$ oc tag <source> <destination>
For example, to configure the
ruby
image streamstatic-2.0
tag to always refer to the current image for theruby
image stream2.0
tag:$ oc tag ruby:2.0 ruby:static-2.0
This creates a new image stream tag named
static-2.0
in theruby
image stream. The new tag directly references the image id that theruby:2.0
image stream tag pointed to at the timeoc tag
was run, and the image it points to never changes.To ensure the destination tag is updated when the source tag changes, use the
--alias=true
flag:$ oc tag --alias=true <source> <destination>
Use a tracking tag for creating permanent aliases, for example, latest
or stable
. The tag only works correctly within a single image stream. Trying to create a cross-image stream alias produces an error.
-
You can also add the
--scheduled=true
flag to have the destination tag be refreshed, or re-imported, periodically. The period is configured globally at the system level. The
--reference
flag creates an image stream tag that is not imported. The tag points to the source location, permanently.If you want to instruct OpenShift Container Platform to always fetch the tagged image from the integrated registry, use
--reference-policy=local
. The registry uses the pull-through feature to serve the image to the client. By default, the image blobs are mirrored locally by the registry. As a result, they can be pulled more quickly the next time they are needed. The flag also allows for pulling from insecure registries without a need to supply--insecure-registry
to the container runtime as long as the image stream has an insecure annotation or the tag has an insecure import policy.
5.2.4. Removing tags from image streams
You can remove tags from an image stream.
Procedure
To remove a tag completely from an image stream run:
$ oc delete istag/ruby:latest
or:
$ oc tag -d ruby:latest
5.2.5. Referencing images in imagestreams
You can use tags to reference images in image streams using the following reference types.
Reference type | Description |
---|---|
|
An |
|
An |
|
A |
When viewing example image stream definitions you may notice they contain definitions of ImageStreamTag
and references to DockerImage
, but nothing related to ImageStreamImage
.
This is because the ImageStreamImage
objects are automatically created in OpenShift Container Platform when you import or tag an image into the image stream. You should never have to explicitly define an ImageStreamImage
object in any image stream definition that you use to create image streams.
Procedure
To reference an image for a given image stream and tag, use
ImageStreamTag
:<image_stream_name>:<tag>
To reference an image for a given image stream and image
sha
ID, useImageStreamImage
:<image_stream_name>@<id>
The
<id>
is an immutable identifier for a specific image, also called a digest.To reference or retrieve an image for a given external registry, use
DockerImage
:openshift/ruby-20-centos7:2.0
NoteWhen no tag is specified, it is assumed the
latest
tag is used.You can also reference a third-party registry:
registry.redhat.io/rhel7:latest
Or an image with a digest:
centos/ruby-22-centos7@sha256:3a335d7d8a452970c5b4054ad7118ff134b3a6b50a2bb6d0c07c746e8986b28e
5.3. Image pull policy
Each container in a pod has a container image. After you have created an image and pushed it to a registry, you can then refer to it in the pod.
5.3.1. Image pull policy overview
When OpenShift Container Platform creates containers, it uses the container imagePullPolicy
to determine if the image should be pulled prior to starting the container. There are three possible values for imagePullPolicy
:
Value | Description |
---|---|
| Always pull the image. |
| Only pull the image if it does not already exist on the node. |
| Never pull the image. |
If a container imagePullPolicy
parameter is not specified, OpenShift Container Platform sets it based on the image tag:
-
If the tag is
latest
, OpenShift Container Platform defaultsimagePullPolicy
toAlways
. -
Otherwise, OpenShift Container Platform defaults
imagePullPolicy
toIfNotPresent
.
5.4. Using image pull secrets
If you are using the OpenShift image registry and are pulling from image streams located in the same project, then your pod service account should already have the correct permissions and no additional action should be required.
However, for other scenarios, such as referencing images across OpenShift Container Platform projects or from secured registries, additional configuration steps are required.
You can obtain the image pull secret from Red Hat OpenShift Cluster Manager. This pull secret is called pullSecret
.
You use this pull secret to authenticate with the services that are provided by the included authorities, Quay.io and registry.redhat.io, which serve the container images for OpenShift Container Platform components.
5.4.1. Allowing pods to reference images across projects
When using the OpenShift image registry, to allow pods in project-a
to reference images in project-b
, a service account in project-a
must be bound to the system:image-puller
role in project-b
.
When you create a pod service account or a namespace, wait until the service account is provisioned with a docker pull secret; if you create a pod before its service account is fully provisioned, the pod fails to access the OpenShift image registry.
Procedure
To allow pods in
project-a
to reference images inproject-b
, bind a service account inproject-a
to thesystem:image-puller
role inproject-b
:$ oc policy add-role-to-user \ system:image-puller system:serviceaccount:project-a:default \ --namespace=project-b
After adding that role, the pods in
project-a
that reference the default service account are able to pull images fromproject-b
.To allow access for any service account in
project-a
, use the group:$ oc policy add-role-to-group \ system:image-puller system:serviceaccounts:project-a \ --namespace=project-b
5.4.2. Allowing pods to reference images from other secured registries
To pull a secured container from other private or secured registries, you must create a pull secret from your container client credentials, such as Docker or Podman, and add it to your service account.
Both Docker and Podman use a configuration file to store authentication details to log in to secured or insecure registry:
-
Docker: By default, Docker uses
$HOME/.docker/config.json
. -
Podman: By default, Podman uses
$HOME/.config/containers/auth.json
.
These files store your authentication information if you have previously logged in to a secured or insecure registry.
Both Docker and Podman credential files and the associated pull secret can contain multiple references to the same registry if they have unique paths, for example, quay.io
and quay.io/<example_repository>
. However, neither Docker nor Podman support multiple entries for the exact same registry path.
Example config.json
file
{ "auths":{ "cloud.openshift.com":{ "auth":"b3Blb=", "email":"you@example.com" }, "quay.io":{ "auth":"b3Blb=", "email":"you@example.com" }, "quay.io/repository-main":{ "auth":"b3Blb=", "email":"you@example.com" } } }
Example pull secret
apiVersion: v1 data: .dockerconfigjson: ewogICAiYXV0aHMiOnsKICAgICAgIm0iOnsKICAgICAgIsKICAgICAgICAgImF1dGgiOiJiM0JsYj0iLAogICAgICAgICAiZW1haWwiOiJ5b3VAZXhhbXBsZS5jb20iCiAgICAgIH0KICAgfQp9Cg== kind: Secret metadata: creationTimestamp: "2021-09-09T19:10:11Z" name: pull-secret namespace: default resourceVersion: "37676" uid: e2851531-01bc-48ba-878c-de96cfe31020 type: Opaque
Procedure
Create a secret from an existing authentication file:
For Docker clients using
.docker/config.json
, enter the following command:$ oc create secret generic <pull_secret_name> \ --from-file=.dockerconfigjson=<path/to/.docker/config.json> \ --type=kubernetes.io/dockerconfigjson
For Podman clients using
.config/containers/auth.json
, enter the following command:$ oc create secret generic <pull_secret_name> \ --from-file=<path/to/.config/containers/auth.json> \ --type=kubernetes.io/podmanconfigjson
If you do not already have a Docker credentials file for the secured registry, you can create a secret by running:
$ oc create secret docker-registry <pull_secret_name> \ --docker-server=<registry_server> \ --docker-username=<user_name> \ --docker-password=<password> \ --docker-email=<email>
To use a secret for pulling images for pods, you must add the secret to your service account. The name of the service account in this example should match the name of the service account the pod uses. The default service account is
default
:$ oc secrets link default <pull_secret_name> --for=pull
5.4.2.1. Pulling from private registries with delegated authentication
A private registry can delegate authentication to a separate service. In these cases, image pull secrets must be defined for both the authentication and registry endpoints.
Procedure
Create a secret for the delegated authentication server:
$ oc create secret docker-registry \ --docker-server=sso.redhat.com \ --docker-username=developer@example.com \ --docker-password=******** \ --docker-email=unused \ redhat-connect-sso secret/redhat-connect-sso
Create a secret for the private registry:
$ oc create secret docker-registry \ --docker-server=privateregistry.example.com \ --docker-username=developer@example.com \ --docker-password=******** \ --docker-email=unused \ private-registry secret/private-registry
5.4.3. Updating the global cluster pull secret
You can update the global pull secret for your cluster by either replacing the current pull secret or appending a new pull secret.
To transfer your cluster to another owner, you must first initiate the transfer in OpenShift Cluster Manager, and then update the pull secret on the cluster. Updating a cluster’s pull secret without initiating the transfer in OpenShift Cluster Manager causes the cluster to stop reporting Telemetry metrics in OpenShift Cluster Manager.
For more information about transferring cluster ownership, see "Transferring cluster ownership" in the Red Hat OpenShift Cluster Manager documentation.
Prerequisites
-
You have access to the cluster as a user with the
cluster-admin
role.
Procedure
Optional: To append a new pull secret to the existing pull secret, complete the following steps:
Enter the following command to download the pull secret:
$ oc get secret/pull-secret -n openshift-config --template='{{index .data ".dockerconfigjson" | base64decode}}' ><pull_secret_location> 1
- 1
- Provide the path to the pull secret file.
Enter the following command to add the new pull secret:
$ oc registry login --registry="<registry>" \ 1 --auth-basic="<username>:<password>" \ 2 --to=<pull_secret_location> 3
Alternatively, you can perform a manual update to the pull secret file.
Enter the following command to update the global pull secret for your cluster:
$ oc set data secret/pull-secret -n openshift-config --from-file=.dockerconfigjson=<pull_secret_location> 1
- 1
- Provide the path to the new pull secret file.
This update is rolled out to all nodes, which can take some time depending on the size of your cluster.
NoteAs of OpenShift Container Platform 4.7.4, changes to the global pull secret no longer trigger a node drain or reboot.
Chapter 6. Managing image streams
Image streams provide a means of creating and updating container images in an on-going way. As improvements are made to an image, tags can be used to assign new version numbers and keep track of changes. This document describes how image streams are managed.
6.1. Why use imagestreams
An image stream and its associated tags provide an abstraction for referencing container images from within OpenShift Container Platform. The image stream and its tags allow you to see what images are available and ensure that you are using the specific image you need even if the image in the repository changes.
Image streams do not contain actual image data, but present a single virtual view of related images, similar to an image repository.
You can configure builds and deployments to watch an image stream for notifications when new images are added and react by performing a build or deployment, respectively.
For example, if a deployment is using a certain image and a new version of that image is created, a deployment could be automatically performed to pick up the new version of the image.
However, if the image stream tag used by the deployment or build is not updated, then even if the container image in the container image registry is updated, the build or deployment continues using the previous, presumably known good image.
The source images can be stored in any of the following:
- OpenShift Container Platform’s integrated registry.
- An external registry, for example registry.redhat.io or quay.io.
- Other image streams in the OpenShift Container Platform cluster.
When you define an object that references an image stream tag, such as a build or deployment configuration, you point to an image stream tag and not the repository. When you build or deploy your application, OpenShift Container Platform queries the repository using the image stream tag to locate the associated ID of the image and uses that exact image.
The image stream metadata is stored in the etcd instance along with other cluster information.
Using image streams has several significant benefits:
- You can tag, rollback a tag, and quickly deal with images, without having to re-push using the command line.
- You can trigger builds and deployments when a new image is pushed to the registry. Also, OpenShift Container Platform has generic triggers for other resources, such as Kubernetes objects.
- You can mark a tag for periodic re-import. If the source image has changed, that change is picked up and reflected in the image stream, which triggers the build or deployment flow, depending upon the build or deployment configuration.
- You can share images using fine-grained access control and quickly distribute images across your teams.
- If the source image changes, the image stream tag still points to a known-good version of the image, ensuring that your application does not break unexpectedly.
- You can configure security around who can view and use the images through permissions on the image stream objects.
- Users that lack permission to read or list images on the cluster level can still retrieve the images tagged in a project using image streams.
6.2. Configuring image streams
An ImageStream
object file contains the following elements.
Imagestream object definition
apiVersion: image.openshift.io/v1 kind: ImageStream metadata: annotations: openshift.io/generated-by: OpenShiftNewApp labels: app: ruby-sample-build template: application-template-stibuild name: origin-ruby-sample 1 namespace: test spec: {} status: dockerImageRepository: 172.30.56.218:5000/test/origin-ruby-sample 2 tags: - items: - created: 2017-09-02T10:15:09Z dockerImageReference: 172.30.56.218:5000/test/origin-ruby-sample@sha256:47463d94eb5c049b2d23b03a9530bf944f8f967a0fe79147dd6b9135bf7dd13d 3 generation: 2 image: sha256:909de62d1f609a717ec433cc25ca5cf00941545c83a01fb31527771e1fab3fc5 4 - created: 2017-09-01T13:40:11Z dockerImageReference: 172.30.56.218:5000/test/origin-ruby-sample@sha256:909de62d1f609a717ec433cc25ca5cf00941545c83a01fb31527771e1fab3fc5 generation: 1 image: sha256:47463d94eb5c049b2d23b03a9530bf944f8f967a0fe79147dd6b9135bf7dd13d tag: latest 5
- 1
- The name of the image stream.
- 2
- Docker repository path where new images can be pushed to add or update them in this image stream.
- 3
- The SHA identifier that this image stream tag currently references. Resources that reference this image stream tag use this identifier.
- 4
- The SHA identifier that this image stream tag previously referenced. Can be used to rollback to an older image.
- 5
- The image stream tag name.
6.3. Image stream images
An image stream image points from within an image stream to a particular image ID.
Image stream images allow you to retrieve metadata about an image from a particular image stream where it is tagged.
Image stream image objects are automatically created in OpenShift Container Platform whenever you import or tag an image into the image stream. You should never have to explicitly define an image stream image object in any image stream definition that you use to create image streams.
The image stream image consists of the image stream name and image ID from the repository, delimited by an @
sign:
<image-stream-name>@<image-id>
To refer to the image in the ImageStream
object example, the image stream image looks like:
origin-ruby-sample@sha256:47463d94eb5c049b2d23b03a9530bf944f8f967a0fe79147dd6b9135bf7dd13d
6.4. Image stream tags
An image stream tag is a named pointer to an image in an image stream. It is abbreviated as istag
. An image stream tag is used to reference or retrieve an image for a given image stream and tag.
Image stream tags can reference any local or externally managed image. It contains a history of images represented as a stack of all images the tag ever pointed to. Whenever a new or existing image is tagged under particular image stream tag, it is placed at the first position in the history stack. The image previously occupying the top position is available at the second position. This allows for easy rollbacks to make tags point to historical images again.
The following image stream tag is from an ImageStream
object:
Image stream tag with two images in its history
kind: ImageStream apiVersion: image.openshift.io/v1 metadata: name: my-image-stream # ... tags: - items: - created: 2017-09-02T10:15:09Z dockerImageReference: 172.30.56.218:5000/test/origin-ruby-sample@sha256:47463d94eb5c049b2d23b03a9530bf944f8f967a0fe79147dd6b9135bf7dd13d generation: 2 image: sha256:909de62d1f609a717ec433cc25ca5cf00941545c83a01fb31527771e1fab3fc5 - created: 2017-09-01T13:40:11Z dockerImageReference: 172.30.56.218:5000/test/origin-ruby-sample@sha256:909de62d1f609a717ec433cc25ca5cf00941545c83a01fb31527771e1fab3fc5 generation: 1 image: sha256:47463d94eb5c049b2d23b03a9530bf944f8f967a0fe79147dd6b9135bf7dd13d tag: latest # ...
Image stream tags can be permanent tags or tracking tags.
- Permanent tags are version-specific tags that point to a particular version of an image, such as Python 3.5.
Tracking tags are reference tags that follow another image stream tag and can be updated to change which image they follow, like a symlink. These new levels are not guaranteed to be backwards-compatible.
For example, the
latest
image stream tags that ship with OpenShift Container Platform are tracking tags. This means consumers of thelatest
image stream tag are updated to the newest level of the framework provided by the image when a new level becomes available. Alatest
image stream tag tov3.10
can be changed tov3.11
at any time. It is important to be aware that theselatest
image stream tags behave differently than the Dockerlatest
tag. Thelatest
image stream tag, in this case, does not point to the latest image in the Docker repository. It points to another image stream tag, which might not be the latest version of an image. For example, if thelatest
image stream tag points tov3.10
of an image, when the3.11
version is released, thelatest
tag is not automatically updated tov3.11
, and remains atv3.10
until it is manually updated to point to av3.11
image stream tag.NoteTracking tags are limited to a single image stream and cannot reference other image streams.
You can create your own image stream tags for your own needs.
The image stream tag is composed of the name of the image stream and a tag, separated by a colon:
<imagestream name>:<tag>
For example, to refer to the sha256:47463d94eb5c049b2d23b03a9530bf944f8f967a0fe79147dd6b9135bf7dd13d
image in the ImageStream
object example earlier, the image stream tag would be:
origin-ruby-sample:latest
6.5. Image stream change triggers
Image stream triggers allow your builds and deployments to be automatically invoked when a new version of an upstream image is available.
For example, builds and deployments can be automatically started when an image stream tag is modified. This is achieved by monitoring that particular image stream tag and notifying the build or deployment when a change is detected.
6.6. Image stream mapping
When the integrated registry receives a new image, it creates and sends an image stream mapping to OpenShift Container Platform, providing the image’s project, name, tag, and image metadata.
Configuring image stream mappings is an advanced feature.
This information is used to create a new image, if it does not already exist, and to tag the image into the image stream. OpenShift Container Platform stores complete metadata about each image, such as commands, entry point, and environment variables. Images in OpenShift Container Platform are immutable and the maximum name length is 63 characters.
The following image stream mapping example results in an image being tagged as test/origin-ruby-sample:latest
:
Image stream mapping object definition
apiVersion: image.openshift.io/v1 kind: ImageStreamMapping metadata: creationTimestamp: null name: origin-ruby-sample namespace: test tag: latest image: dockerImageLayers: - name: sha256:5f70bf18a086007016e948b04aed3b82103a36bea41755b6cddfaf10ace3c6ef size: 0 - name: sha256:ee1dd2cb6df21971f4af6de0f1d7782b81fb63156801cfde2bb47b4247c23c29 size: 196634330 - name: sha256:5f70bf18a086007016e948b04aed3b82103a36bea41755b6cddfaf10ace3c6ef size: 0 - name: sha256:5f70bf18a086007016e948b04aed3b82103a36bea41755b6cddfaf10ace3c6ef size: 0 - name: sha256:ca062656bff07f18bff46be00f40cfbb069687ec124ac0aa038fd676cfaea092 size: 177723024 - name: sha256:63d529c59c92843c395befd065de516ee9ed4995549f8218eac6ff088bfa6b6e size: 55679776 - name: sha256:92114219a04977b5563d7dff71ec4caa3a37a15b266ce42ee8f43dba9798c966 size: 11939149 dockerImageMetadata: Architecture: amd64 Config: Cmd: - /usr/libexec/s2i/run Entrypoint: - container-entrypoint Env: - RACK_ENV=production - OPENSHIFT_BUILD_NAMESPACE=test - OPENSHIFT_BUILD_SOURCE=https://github.com/openshift/ruby-hello-world.git - EXAMPLE=sample-app - OPENSHIFT_BUILD_NAME=ruby-sample-build-1 - PATH=/opt/app-root/src/bin:/opt/app-root/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin - STI_SCRIPTS_URL=image:///usr/libexec/s2i - STI_SCRIPTS_PATH=/usr/libexec/s2i - HOME=/opt/app-root/src - BASH_ENV=/opt/app-root/etc/scl_enable - ENV=/opt/app-root/etc/scl_enable - PROMPT_COMMAND=. /opt/app-root/etc/scl_enable - RUBY_VERSION=2.2 ExposedPorts: 8080/tcp: {} Labels: build-date: 2015-12-23 io.k8s.description: Platform for building and running Ruby 2.2 applications io.k8s.display-name: 172.30.56.218:5000/test/origin-ruby-sample:latest io.openshift.build.commit.author: Ben Parees <bparees@users.noreply.github.com> io.openshift.build.commit.date: Wed Jan 20 10:14:27 2016 -0500 io.openshift.build.commit.id: 00cadc392d39d5ef9117cbc8a31db0889eedd442 io.openshift.build.commit.message: 'Merge pull request #51 from php-coder/fix_url_and_sti' io.openshift.build.commit.ref: master io.openshift.build.image: centos/ruby-22-centos7@sha256:3a335d7d8a452970c5b4054ad7118ff134b3a6b50a2bb6d0c07c746e8986b28e io.openshift.build.source-location: https://github.com/openshift/ruby-hello-world.git io.openshift.builder-base-version: 8d95148 io.openshift.builder-version: 8847438ba06307f86ac877465eadc835201241df io.openshift.s2i.scripts-url: image:///usr/libexec/s2i io.openshift.tags: builder,ruby,ruby22 io.s2i.scripts-url: image:///usr/libexec/s2i license: GPLv2 name: CentOS Base Image vendor: CentOS User: "1001" WorkingDir: /opt/app-root/src Container: 86e9a4a3c760271671ab913616c51c9f3cea846ca524bf07c04a6f6c9e103a76 ContainerConfig: AttachStdout: true Cmd: - /bin/sh - -c - tar -C /tmp -xf - && /usr/libexec/s2i/assemble Entrypoint: - container-entrypoint Env: - RACK_ENV=production - OPENSHIFT_BUILD_NAME=ruby-sample-build-1 - OPENSHIFT_BUILD_NAMESPACE=test - OPENSHIFT_BUILD_SOURCE=https://github.com/openshift/ruby-hello-world.git - EXAMPLE=sample-app - PATH=/opt/app-root/src/bin:/opt/app-root/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin - STI_SCRIPTS_URL=image:///usr/libexec/s2i - STI_SCRIPTS_PATH=/usr/libexec/s2i - HOME=/opt/app-root/src - BASH_ENV=/opt/app-root/etc/scl_enable - ENV=/opt/app-root/etc/scl_enable - PROMPT_COMMAND=. /opt/app-root/etc/scl_enable - RUBY_VERSION=2.2 ExposedPorts: 8080/tcp: {} Hostname: ruby-sample-build-1-build Image: centos/ruby-22-centos7@sha256:3a335d7d8a452970c5b4054ad7118ff134b3a6b50a2bb6d0c07c746e8986b28e OpenStdin: true StdinOnce: true User: "1001" WorkingDir: /opt/app-root/src Created: 2016-01-29T13:40:00Z DockerVersion: 1.8.2.fc21 Id: 9d7fd5e2d15495802028c569d544329f4286dcd1c9c085ff5699218dbaa69b43 Parent: 57b08d979c86f4500dc8cad639c9518744c8dd39447c055a3517dc9c18d6fccd Size: 441976279 apiVersion: "1.0" kind: DockerImage dockerImageMetadataVersion: "1.0" dockerImageReference: 172.30.56.218:5000/test/origin-ruby-sample@sha256:47463d94eb5c049b2d23b03a9530bf944f8f967a0fe79147dd6b9135bf7dd13d
6.7. Working with image streams
The following sections describe how to use image streams and image stream tags.
Do not run workloads in or share access to default projects. Default projects are reserved for running core cluster components.
The following default projects are considered highly privileged: default
, kube-public
, kube-system
, openshift
, openshift-infra
, openshift-node
, and other system-created projects that have the openshift.io/run-level
label set to 0
or 1
. Functionality that relies on admission plugins, such as pod security admission, security context constraints, cluster resource quotas, and image reference resolution, does not work in highly privileged projects.
6.7.1. Getting information about image streams
You can get general information about the image stream and detailed information about all the tags it is pointing to.
Procedure
To get general information about the image stream and detailed information about all the tags it is pointing to, enter the following command:
$ oc describe is/<image-name>
For example:
$ oc describe is/python
Example output
Name: python Namespace: default Created: About a minute ago Labels: <none> Annotations: openshift.io/image.dockerRepositoryCheck=2017-10-02T17:05:11Z Docker Pull Spec: docker-registry.default.svc:5000/default/python Image Lookup: local=false Unique Images: 1 Tags: 1 3.5 tagged from centos/python-35-centos7 * centos/python-35-centos7@sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25 About a minute ago
To get all of the information available about a particular image stream tag, enter the following command:
$ oc describe istag/<image-stream>:<tag-name>
For example:
$ oc describe istag/python:latest
Example output
Image Name: sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25 Docker Image: centos/python-35-centos7@sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25 Name: sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25 Created: 2 minutes ago Image Size: 251.2 MB (first layer 2.898 MB, last binary layer 72.26 MB) Image Created: 2 weeks ago Author: <none> Arch: amd64 Entrypoint: container-entrypoint Command: /bin/sh -c $STI_SCRIPTS_PATH/usage Working Dir: /opt/app-root/src User: 1001 Exposes Ports: 8080/tcp Docker Labels: build-date=20170801
NoteMore information is output than shown.
Enter the following command to discover which architecture or operating system that an image stream tag supports:
$ oc get istag <image-stream-tag> -ojsonpath="{range .image.dockerImageManifests[*]}{.os}/{.architecture}{'\n'}{end}"
For example:
$ oc get istag busybox:latest -ojsonpath="{range .image.dockerImageManifests[*]}{.os}/{.architecture}{'\n'}{end}"
Example output
linux/amd64 linux/arm linux/arm64 linux/386 linux/mips64le linux/ppc64le linux/riscv64 linux/s390x
6.7.2. Adding tags to an image stream
You can add additional tags to image streams.
Procedure
Add a tag that points to one of the existing tags by using the `oc tag`command:
$ oc tag <image-name:tag1> <image-name:tag2>
For example:
$ oc tag python:3.5 python:latest
Example output
Tag python:latest set to python@sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25.
Confirm the image stream has two tags, one,
3.5
, pointing at the external container image and another tag,latest
, pointing to the same image because it was created based on the first tag.$ oc describe is/python
Example output
Name: python Namespace: default Created: 5 minutes ago Labels: <none> Annotations: openshift.io/image.dockerRepositoryCheck=2017-10-02T17:05:11Z Docker Pull Spec: docker-registry.default.svc:5000/default/python Image Lookup: local=false Unique Images: 1 Tags: 2 latest tagged from python@sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25 * centos/python-35-centos7@sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25 About a minute ago 3.5 tagged from centos/python-35-centos7 * centos/python-35-centos7@sha256:49c18358df82f4577386404991c51a9559f243e0b1bdc366df25 5 minutes ago
6.7.3. Adding tags for an external image
You can add tags for external images.
Procedure
Add tags pointing to internal or external images, by using the
oc tag
command for all tag-related operations:$ oc tag <repository/image> <image-name:tag>
For example, this command maps the
docker.io/python:3.6.0
image to the3.6
tag in thepython
image stream.$ oc tag docker.io/python:3.6.0 python:3.6
Example output
Tag python:3.6 set to docker.io/python:3.6.0.
If the external image is secured, you must create a secret with credentials for accessing that registry.
6.7.4. Updating image stream tags
You can update a tag to reflect another tag in an image stream.
Procedure
Update a tag:
$ oc tag <image-name:tag> <image-name:latest>
For example, the following updates the
latest
tag to reflect the3.6
tag in an image stream:$ oc tag python:3.6 python:latest
Example output
Tag python:latest set to python@sha256:438208801c4806548460b27bd1fbcb7bb188273d13871ab43f.
6.7.5. Removing image stream tags
You can remove old tags from an image stream.
Procedure
Remove old tags from an image stream:
$ oc tag -d <image-name:tag>
For example:
$ oc tag -d python:3.6
Example output
Deleted tag default/python:3.6
See Removing deprecated image stream tags from the Cluster Samples Operator for more information on how the Cluster Samples Operator handles deprecated image stream tags.
6.7.6. Configuring periodic importing of image stream tags
When working with an external container image registry, to periodically re-import an image, for example to get latest security updates, you can use the --scheduled
flag.
Procedure
Schedule importing images:
$ oc tag <repository/image> <image-name:tag> --scheduled
For example:
$ oc tag docker.io/python:3.6.0 python:3.6 --scheduled
Example output
Tag python:3.6 set to import docker.io/python:3.6.0 periodically.
This command causes OpenShift Container Platform to periodically update this particular image stream tag. This period is a cluster-wide setting set to 15 minutes by default.
Remove the periodic check, re-run above command but omit the
--scheduled
flag. This will reset its behavior to default.$ oc tag <repositiory/image> <image-name:tag>
6.8. Importing and working with images and image streams
The following sections describe how to import, and work with, image streams.
6.8.1. Importing images and image streams from private registries
An image stream can be configured to import tag and image metadata from private image registries requiring authentication. This procedures applies if you change the registry that the Cluster Samples Operator uses to pull content from to something other than registry.redhat.io.
When importing from insecure or secure registries, the registry URL defined in the secret must include the :80
port suffix or the secret is not used when attempting to import from the registry.
Procedure
You must create a
secret
object that is used to store your credentials by entering the following command:$ oc create secret generic <secret_name> --from-file=.dockerconfigjson=<file_absolute_path> --type=kubernetes.io/dockerconfigjson
After the secret is configured, create the new image stream or enter the
oc import-image
command:$ oc import-image <imagestreamtag> --from=<image> --confirm
During the import process, OpenShift Container Platform picks up the secrets and provides them to the remote party.
6.8.1.1. Allowing pods to reference images from other secured registries
To pull a secured container from other private or secured registries, you must create a pull secret from your container client credentials, such as Docker or Podman, and add it to your service account.
Both Docker and Podman use a configuration file to store authentication details to log in to secured or insecure registry:
-
Docker: By default, Docker uses
$HOME/.docker/config.json
. -
Podman: By default, Podman uses
$HOME/.config/containers/auth.json
.
These files store your authentication information if you have previously logged in to a secured or insecure registry.
Both Docker and Podman credential files and the associated pull secret can contain multiple references to the same registry if they have unique paths, for example, quay.io
and quay.io/<example_repository>
. However, neither Docker nor Podman support multiple entries for the exact same registry path.
Example config.json
file
{ "auths":{ "cloud.openshift.com":{ "auth":"b3Blb=", "email":"you@example.com" }, "quay.io":{ "auth":"b3Blb=", "email":"you@example.com" }, "quay.io/repository-main":{ "auth":"b3Blb=", "email":"you@example.com" } } }
Example pull secret
apiVersion: v1 data: .dockerconfigjson: ewogICAiYXV0aHMiOnsKICAgICAgIm0iOnsKICAgICAgIsKICAgICAgICAgImF1dGgiOiJiM0JsYj0iLAogICAgICAgICAiZW1haWwiOiJ5b3VAZXhhbXBsZS5jb20iCiAgICAgIH0KICAgfQp9Cg== kind: Secret metadata: creationTimestamp: "2021-09-09T19:10:11Z" name: pull-secret namespace: default resourceVersion: "37676" uid: e2851531-01bc-48ba-878c-de96cfe31020 type: Opaque
Procedure
Create a secret from an existing authentication file:
For Docker clients using
.docker/config.json
, enter the following command:$ oc create secret generic <pull_secret_name> \ --from-file=.dockerconfigjson=<path/to/.docker/config.json> \ --type=kubernetes.io/dockerconfigjson
For Podman clients using
.config/containers/auth.json
, enter the following command:$ oc create secret generic <pull_secret_name> \ --from-file=<path/to/.config/containers/auth.json> \ --type=kubernetes.io/podmanconfigjson
If you do not already have a Docker credentials file for the secured registry, you can create a secret by running:
$ oc create secret docker-registry <pull_secret_name> \ --docker-server=<registry_server> \ --docker-username=<user_name> \ --docker-password=<password> \ --docker-email=<email>
To use a secret for pulling images for pods, you must add the secret to your service account. The name of the service account in this example should match the name of the service account the pod uses. The default service account is
default
:$ oc secrets link default <pull_secret_name> --for=pull
6.8.2. Working with manifest lists
You can import a single sub-manifest, or all manifests, of a manifest list when using oc import-image
or oc tag
CLI commands by adding the --import-mode
flag.
Refer to the commands below to create an image stream that includes a single sub-manifest or multi-architecture images.
Procedure
Create an image stream that includes multi-architecture images, and sets the import mode to
PreserveOriginal
, by entering the following command:$ oc import-image <multiarch-image-stream-tag> --from=<registry>/<project_name>/<image-name> \ --import-mode='PreserveOriginal' --reference-policy=local --confirm
Example output
--- Arch: <none> Manifests: linux/amd64 sha256:6e325b86566fafd3c4683a05a219c30c421fbccbf8d87ab9d20d4ec1131c3451 linux/arm64 sha256:d8fad562ffa75b96212c4a6dc81faf327d67714ed85475bf642729703a2b5bf6 linux/ppc64le sha256:7b7e25338e40d8bdeb1b28e37fef5e64f0afd412530b257f5b02b30851f416e1 ---
Alternatively, enter the following command to import an image with the
Legacy
import mode, which discards manifest lists and imports a single sub-manifest:$ oc import-image <multiarch-image-stream-tag> --from=<registry>/<project_name>/<image-name> \ --import-mode='Legacy' --confirm
NoteThe
--import-mode=
default value isLegacy
. Excluding this value, or failing to specify eitherLegacy
orPreserveOriginal
, imports a single sub-manifest. An invalid import mode returns the following error:error: valid ImportMode values are Legacy or PreserveOriginal
.
Limitations
Working with manifest lists has the following limitations:
In some cases, users might want to use sub-manifests directly. When
oc adm prune images
is run, or theCronJob
pruner runs, they cannot detect when a sub-manifest list is used. As a result, an administrator usingoc adm prune images
, or theCronJob
pruner, might delete entire manifest lists, including sub-manifests.To avoid this limitation, you can use the manifest list by tag or by digest instead.
6.8.2.1. Configuring periodic importing of manifest lists
To periodically re-import a manifest list, you can use the --scheduled
flag.
Procedure
Set the image stream to periodically update the manifest list by entering the following command:
$ oc import-image <multiarch-image-stream-tag> --from=<registry>/<project_name>/<image-name> \ --import-mode='PreserveOriginal' --scheduled=true
6.8.2.2. Configuring SSL/TSL when importing manifest lists
To configure SSL/TSL when importing a manifest list, you can use the --insecure
flag.
Procedure
Set
--insecure=true
so that importing a manifest list skips SSL/TSL verification. For example:$ oc import-image <multiarch-image-stream-tag> --from=<registry>/<project_name>/<image-name> \ --import-mode='PreserveOriginal' --insecure=true
6.8.3. Specifying architecture for --import-mode
You can swap your imported image stream between multi-architecture and single architecture by excluding or including the --import-mode=
flag
Procedure
Run the following command to update your image stream from multi-architecture to single architecture by excluding the
--import-mode=
flag:$ oc import-image <multiarch-image-stream-tag> --from=<registry>/<project_name>/<image-name>
Run the following command to update your image stream from single-architecture to multi-architecture:
$ oc import-image <multiarch-image-stream-tag> --from=<registry>/<project_name>/<image-name> \ --import-mode='PreserveOriginal'
6.8.4. Configuration fields for --import-mode
The following table describes the options available for the --import-mode=
flag:
Parameter | Description |
---|---|
Legacy |
The default option for
|
PreserveOriginal | When specified, the original manifest is preserved. For manifest lists, the manifest list and all of its sub-manifests are imported. |
Chapter 7. Using image streams with Kubernetes resources
Image streams, being OpenShift Container Platform native resources, work with all native resources available in OpenShift Container Platform, such as Build
or DeploymentConfigs
resources. It is also possible to make them work with native Kubernetes resources, such as Job
, ReplicationController
, ReplicaSet
or Kubernetes Deployment
resources.
7.1. Enabling image streams with Kubernetes resources
When using image streams with Kubernetes resources, you can only reference image streams that reside in the same project as the resource. The image stream reference must consist of a single segment value, for example ruby:2.5
, where ruby
is the name of an image stream that has a tag named 2.5
and resides in the same project as the resource making the reference.
Do not run workloads in or share access to default projects. Default projects are reserved for running core cluster components.
The following default projects are considered highly privileged: default
, kube-public
, kube-system
, openshift
, openshift-infra
, openshift-node
, and other system-created projects that have the openshift.io/run-level
label set to 0
or 1
. Functionality that relies on admission plugins, such as pod security admission, security context constraints, cluster resource quotas, and image reference resolution, does not work in highly privileged projects.
There are two ways to enable image streams with Kubernetes resources:
- Enabling image stream resolution on a specific resource. This allows only this resource to use the image stream name in the image field.
- Enabling image stream resolution on an image stream. This allows all resources pointing to this image stream to use it in the image field.
Procedure
You can use oc set image-lookup
to enable image stream resolution on a specific resource or image stream resolution on an image stream.
To allow all resources to reference the image stream named
mysql
, enter the following command:$ oc set image-lookup mysql
This sets the
Imagestream.spec.lookupPolicy.local
field to true.Imagestream with image lookup enabled
apiVersion: image.openshift.io/v1 kind: ImageStream metadata: annotations: openshift.io/display-name: mysql name: mysql namespace: myproject spec: lookupPolicy: local: true
When enabled, the behavior is enabled for all tags within the image stream.
Then you can query the image streams and see if the option is set:
$ oc set image-lookup imagestream --list
You can enable image lookup on a specific resource.
To allow the Kubernetes deployment named
mysql
to use image streams, run the following command:$ oc set image-lookup deploy/mysql
This sets the
alpha.image.policy.openshift.io/resolve-names
annotation on the deployment.Deployment with image lookup enabled
apiVersion: apps/v1 kind: Deployment metadata: name: mysql namespace: myproject spec: replicas: 1 template: metadata: annotations: alpha.image.policy.openshift.io/resolve-names: '*' spec: containers: - image: mysql:latest imagePullPolicy: Always name: mysql
You can disable image lookup.
To disable image lookup, pass
--enabled=false
:$ oc set image-lookup deploy/mysql --enabled=false
Chapter 8. Triggering updates on image stream changes
When an image stream tag is updated to point to a new image, OpenShift Container Platform can automatically take action to roll the new image out to resources that were using the old image. You configure this behavior in different ways depending on the type of resource that references the image stream tag.
8.1. OpenShift Container Platform resources
OpenShift Container Platform deployment configurations and build configurations can be automatically triggered by changes to image stream tags. The triggered action can be run using the new value of the image referenced by the updated image stream tag.
8.2. Triggering Kubernetes resources
Kubernetes resources do not have fields for triggering, unlike deployment and build configurations, which include as part of their API definition a set of fields for controlling triggers. Instead, you can use annotations in OpenShift Container Platform to request triggering.
The annotation is defined as follows:
apiVersion: v1 kind: Pod metadata: annotations: image.openshift.io/triggers: [ { "from": { "kind": "ImageStreamTag", 1 "name": "example:latest", 2 "namespace": "myapp" 3 }, "fieldPath": "spec.template.spec.containers[?(@.name==\"web\")].image", 4 "paused": false 5 }, # ... ] # ...
- 1
- Required:
kind
is the resource to trigger from must beImageStreamTag
. - 2
- Required:
name
must be the name of an image stream tag. - 3
- Optional:
namespace
defaults to the namespace of the object. - 4
- Required:
fieldPath
is the JSON path to change. This field is limited and accepts only a JSON path expression that precisely matches a container by ID or index. For pods, the JSON path isspec.containers[?(@.name='web')].image
. - 5
- Optional:
paused
is whether or not the trigger is paused, and the default value isfalse
. Setpaused
totrue
to temporarily disable this trigger.
When one of the core Kubernetes resources contains both a pod template and this annotation, OpenShift Container Platform attempts to update the object by using the image currently associated with the image stream tag that is referenced by trigger. The update is performed against the fieldPath
specified.
Examples of core Kubernetes resources that can contain both a pod template and annotation include:
-
CronJobs
-
Deployments
-
StatefulSets
-
DaemonSets
-
Jobs
-
ReplicationControllers
-
Pods
8.3. Setting the image trigger on Kubernetes resources
When adding an image trigger to deployments, you can use the oc set triggers
command. For example, the sample command in this procedure adds an image change trigger to the deployment named example
so that when the example:latest
image stream tag is updated, the web
container inside the deployment updates with the new image value. This command sets the correct image.openshift.io/triggers
annotation on the deployment resource.
Procedure
Trigger Kubernetes resources by entering the
oc set triggers
command:$ oc set triggers deploy/example --from-image=example:latest -c web
Example deployment with trigger annotation
apiVersion: apps/v1 kind: Deployment metadata: annotations: image.openshift.io/triggers: '[{"from":{"kind":"ImageStreamTag","name":"example:latest"},"fieldPath":"spec.template.spec.containers[?(@.name==\"container\")].image"}]' # ...
Unless the deployment is paused, this pod template update automatically causes a deployment to occur with the new image value.
Chapter 9. Image configuration resources
Use the following procedure to configure image registries.
9.1. Image controller configuration parameters
The image.config.openshift.io/cluster
resource holds cluster-wide information about how to handle images. The canonical, and only valid name is cluster
. Its spec
offers the following configuration parameters.
Parameters such as DisableScheduledImport
, MaxImagesBulkImportedPerRepository
, MaxScheduledImportsPerMinute
, ScheduledImageImportMinimumIntervalSeconds
, InternalRegistryHostname
are not configurable.
Parameter | Description |
---|---|
|
Limits the container image registries from which normal users can import images. Set this list to the registries that you trust to contain valid images, and that you want applications to be able to import from. Users with permission to create images or Every element of this list contains a location of the registry specified by the registry domain name.
|
|
A reference to a config map containing additional CAs that should be trusted during
The namespace for this config map is |
|
Provides the hostnames for the default external image registry. The external hostname should be set only when the image registry is exposed externally. The first value is used in |
| Contains configuration that determines how the container runtime should treat individual registries when accessing images for builds and pods. For instance, whether or not to allow insecure access. It does not contain configuration for the internal cluster registry.
Either |
When the allowedRegistries
parameter is defined, all registries, including registry.redhat.io
and quay.io
registries and the default OpenShift image registry, are blocked unless explicitly listed. When using the parameter, to prevent pod failure, add all registries including the registry.redhat.io
and quay.io
registries and the internalRegistryHostname
to the allowedRegistries
list, as they are required by payload images within your environment. For disconnected clusters, mirror registries should also be added.
The status
field of the image.config.openshift.io/cluster
resource holds observed values from the cluster.
Parameter | Description |
---|---|
|
Set by the Image Registry Operator, which controls the |
|
Set by the Image Registry Operator, provides the external hostnames for the image registry when it is exposed externally. The first value is used in |
9.2. Configuring image registry settings
You can configure image registry settings by editing the image.config.openshift.io/cluster
custom resource (CR). When changes to the registry are applied to the image.config.openshift.io/cluster
CR, the Machine Config Operator (MCO) performs the following sequential actions:
- Cordons the node
- Applies changes by restarting CRI-O
Uncordons the node
NoteThe MCO does not restart nodes when it detects changes.
Procedure
Edit the
image.config.openshift.io/cluster
custom resource:$ oc edit image.config.openshift.io/cluster
The following is an example
image.config.openshift.io/cluster
CR:apiVersion: config.openshift.io/v1 kind: Image 1 metadata: annotations: release.openshift.io/create-only: "true" creationTimestamp: "2019-05-17T13:44:26Z" generation: 1 name: cluster resourceVersion: "8302" selfLink: /apis/config.openshift.io/v1/images/cluster uid: e34555da-78a9-11e9-b92b-06d6c7da38dc spec: allowedRegistriesForImport: 2 - domainName: quay.io insecure: false additionalTrustedCA: 3 name: myconfigmap registrySources: 4 allowedRegistries: - example.com - quay.io - registry.redhat.io - image-registry.openshift-image-registry.svc:5000 - reg1.io/myrepo/myapp:latest insecureRegistries: - insecure.com status: internalRegistryHostname: image-registry.openshift-image-registry.svc:5000
- 1
Image
: Holds cluster-wide information about how to handle images. The canonical, and only valid name iscluster
.- 2
allowedRegistriesForImport
: Limits the container image registries from which normal users may import images. Set this list to the registries that you trust to contain valid images, and that you want applications to be able to import from. Users with permission to create images orImageStreamMappings
from the API are not affected by this policy. Typically only cluster administrators have the appropriate permissions.- 3
additionalTrustedCA
: A reference to a config map containing additional certificate authorities (CA) that are trusted during image stream import, pod image pull,openshift-image-registry
pullthrough, and builds. The namespace for this config map isopenshift-config
. The format of the config map is to use the registry hostname as the key, and the PEM certificate as the value, for each additional registry CA to trust.- 4
registrySources
: Contains configuration that determines whether the container runtime allows or blocks individual registries when accessing images for builds and pods. Either theallowedRegistries
parameter or theblockedRegistries
parameter can be set, but not both. You can also define whether or not to allow access to insecure registries or registries that allow registries that use image short names. This example uses theallowedRegistries
parameter, which defines the registries that are allowed to be used. The insecure registryinsecure.com
is also allowed. TheregistrySources
parameter does not contain configuration for the internal cluster registry.
NoteWhen the
allowedRegistries
parameter is defined, all registries, including the registry.redhat.io and quay.io registries and the default OpenShift image registry, are blocked unless explicitly listed. If you use the parameter, to prevent pod failure, you must add theregistry.redhat.io
andquay.io
registries and theinternalRegistryHostname
to theallowedRegistries
list, as they are required by payload images within your environment. Do not add theregistry.redhat.io
andquay.io
registries to theblockedRegistries
list.When using the
allowedRegistries
,blockedRegistries
, orinsecureRegistries
parameter, you can specify an individual repository within a registry. For example:reg1.io/myrepo/myapp:latest
.Insecure external registries should be avoided to reduce possible security risks.
To check that the changes are applied, list your nodes:
$ oc get nodes
Example output
NAME STATUS ROLES AGE VERSION ip-10-0-137-182.us-east-2.compute.internal Ready,SchedulingDisabled worker 65m v1.28.5 ip-10-0-139-120.us-east-2.compute.internal Ready,SchedulingDisabled control-plane 74m v1.28.5 ip-10-0-176-102.us-east-2.compute.internal Ready control-plane 75m v1.28.5 ip-10-0-188-96.us-east-2.compute.internal Ready worker 65m v1.28.5 ip-10-0-200-59.us-east-2.compute.internal Ready worker 63m v1.28.5 ip-10-0-223-123.us-east-2.compute.internal Ready control-plane 73m v1.28.5
9.2.1. Adding specific registries
You can add a list of registries, and optionally an individual repository within a registry, that are permitted for image pull and push actions by editing the image.config.openshift.io/cluster
custom resource (CR). OpenShift Container Platform applies the changes to this CR to all nodes in the cluster.
When pulling or pushing images, the container runtime searches the registries listed under the registrySources
parameter in the image.config.openshift.io/cluster
CR. If you created a list of registries under the allowedRegistries
parameter, the container runtime searches only those registries. Registries not in the list are blocked.
When the allowedRegistries
parameter is defined, all registries, including the registry.redhat.io
and quay.io
registries and the default OpenShift image registry, are blocked unless explicitly listed. If you use the parameter, to prevent pod failure, add the registry.redhat.io
and quay.io
registries and the internalRegistryHostname
to the allowedRegistries
list, as they are required by payload images within your environment. For disconnected clusters, mirror registries should also be added.
Procedure
Edit the
image.config.openshift.io/cluster
custom resource:$ oc edit image.config.openshift.io/cluster
The following is an example
image.config.openshift.io/cluster
CR with an allowed list:apiVersion: config.openshift.io/v1 kind: Image metadata: annotations: release.openshift.io/create-only: "true" creationTimestamp: "2019-05-17T13:44:26Z" generation: 1 name: cluster resourceVersion: "8302" selfLink: /apis/config.openshift.io/v1/images/cluster uid: e34555da-78a9-11e9-b92b-06d6c7da38dc spec: registrySources: 1 allowedRegistries: 2 - example.com - quay.io - registry.redhat.io - reg1.io/myrepo/myapp:latest - image-registry.openshift-image-registry.svc:5000 status: internalRegistryHostname: image-registry.openshift-image-registry.svc:5000
- 1
- Contains configurations that determine how the container runtime should treat individual registries when accessing images for builds and pods. It does not contain configuration for the internal cluster registry.
- 2
- Specify registries, and optionally a repository in that registry, to use for image pull and push actions. All other registries are blocked.
NoteEither the
allowedRegistries
parameter or theblockedRegistries
parameter can be set, but not both.The Machine Config Operator (MCO) watches the
image.config.openshift.io/cluster
resource for any changes to the registries. When the MCO detects a change, it triggers a rollout on nodes in machine config pool (MCP). The allowed registries list is used to update the image signature policy in the/etc/containers/policy.json
file on each node. Changes to the/etc/containers/policy.json
file do not require the node to drain.
Verification
Enter the following command to obtain a list of your nodes:
$ oc get nodes
Example output
NAME STATUS ROLES AGE VERSION <node_name> Ready control-plane,master 37m v1.27.8+4fab27b
Run the following command to enter debug mode on the node:
$ oc debug node/<node_name>
When prompted, enter
chroot /host
into the terminal:sh-4.4# chroot /host
Enter the following command to check that the registries have been added to the policy file:
sh-5.1# cat /etc/containers/policy.json | jq '.'
The following policy indicates that only images from the example.com, quay.io, and registry.redhat.io registries are permitted for image pulls and pushes:
Example 9.1. Example image signature policy file
{ "default":[ { "type":"reject" } ], "transports":{ "atomic":{ "example.com":[ { "type":"insecureAcceptAnything" } ], "image-registry.openshift-image-registry.svc:5000":[ { "type":"insecureAcceptAnything" } ], "insecure.com":[ { "type":"insecureAcceptAnything" } ], "quay.io":[ { "type":"insecureAcceptAnything" } ], "reg4.io/myrepo/myapp:latest":[ { "type":"insecureAcceptAnything" } ], "registry.redhat.io":[ { "type":"insecureAcceptAnything" } ] }, "docker":{ "example.com":[ { "type":"insecureAcceptAnything" } ], "image-registry.openshift-image-registry.svc:5000":[ { "type":"insecureAcceptAnything" } ], "insecure.com":[ { "type":"insecureAcceptAnything" } ], "quay.io":[ { "type":"insecureAcceptAnything" } ], "reg4.io/myrepo/myapp:latest":[ { "type":"insecureAcceptAnything" } ], "registry.redhat.io":[ { "type":"insecureAcceptAnything" } ] }, "docker-daemon":{ "":[ { "type":"insecureAcceptAnything" } ] } } }
If your cluster uses the registrySources.insecureRegistries
parameter, ensure that any insecure registries are included in the allowed list.
For example:
spec: registrySources: insecureRegistries: - insecure.com allowedRegistries: - example.com - quay.io - registry.redhat.io - insecure.com - image-registry.openshift-image-registry.svc:5000
9.2.2. Blocking specific registries
You can block any registry, and optionally an individual repository within a registry, by editing the image.config.openshift.io/cluster
custom resource (CR). OpenShift Container Platform applies the changes to this CR to all nodes in the cluster.
When pulling or pushing images, the container runtime searches the registries listed under the registrySources
parameter in the image.config.openshift.io/cluster
CR. If you created a list of registries under the blockedRegistries
parameter, the container runtime does not search those registries. All other registries are allowed.
To prevent pod failure, do not add the registry.redhat.io
and quay.io
registries to the blockedRegistries
list, as they are required by payload images within your environment.
Procedure
Edit the
image.config.openshift.io/cluster
custom resource:$ oc edit image.config.openshift.io/cluster
The following is an example
image.config.openshift.io/cluster
CR with a blocked list:apiVersion: config.openshift.io/v1 kind: Image metadata: annotations: release.openshift.io/create-only: "true" creationTimestamp: "2019-05-17T13:44:26Z" generation: 1 name: cluster resourceVersion: "8302" selfLink: /apis/config.openshift.io/v1/images/cluster uid: e34555da-78a9-11e9-b92b-06d6c7da38dc spec: registrySources: 1 blockedRegistries: 2 - untrusted.com - reg1.io/myrepo/myapp:latest status: internalRegistryHostname: image-registry.openshift-image-registry.svc:5000
- 1
- Contains configurations that determine how the container runtime should treat individual registries when accessing images for builds and pods. It does not contain configuration for the internal cluster registry.
- 2
- Specify registries, and optionally a repository in that registry, that should not be used for image pull and push actions. All other registries are allowed.
NoteEither the
blockedRegistries
registry or theallowedRegistries
registry can be set, but not both.The Machine Config Operator (MCO) watches the
image.config.openshift.io/cluster
resource for any changes to the registries. When the MCO detects a change, it drains the nodes, applies the change, and uncordons the nodes. After the nodes return to theReady
state, changes to the blocked registries appear in the/etc/containers/registries.conf
file on each node.
Verification
Enter the following command to obtain a list of your nodes:
$ oc get nodes
Example output
NAME STATUS ROLES AGE VERSION <node_name> Ready control-plane,master 37m v1.27.8+4fab27b
Run the following command to enter debug mode on the node:
$ oc debug node/<node_name>
When prompted, enter
chroot /host
into the terminal:sh-4.4# chroot /host
Enter the following command to check that the registries have been added to the policy file:
sh-5.1# cat etc/containers/registries.conf
The following example indicates that images from the
untrusted.com
registry are prevented for image pulls and pushes:Example output
unqualified-search-registries = ["registry.access.redhat.com", "docker.io"] [[registry]] prefix = "" location = "untrusted.com" blocked = true
9.2.2.1. Blocking a payload registry
In a mirroring configuration, you can block upstream payload registries in a disconnected environment using a ImageContentSourcePolicy
(ICSP) object. The following example procedure demonstrates how to block the quay.io/openshift-payload
payload registry.
Procedure
Create the mirror configuration using an
ImageContentSourcePolicy
(ICSP) object to mirror the payload to a registry in your instance. The following example ICSP file mirrors the payloadinternal-mirror.io/openshift-payload
:apiVersion: operator.openshift.io/v1alpha1 kind: ImageContentSourcePolicy metadata: name: my-icsp spec: repositoryDigestMirrors: - mirrors: - internal-mirror.io/openshift-payload source: quay.io/openshift-payload
After the object deploys onto your nodes, verify that the mirror configuration is set by checking the
/etc/containers/registries.conf
file:Example output
[[registry]] prefix = "" location = "quay.io/openshift-payload" mirror-by-digest-only = true [[registry.mirror]] location = "internal-mirror.io/openshift-payload"
Use the following command to edit the
image.config.openshift.io
custom resource file:$ oc edit image.config.openshift.io cluster
To block the payload registry, add the following configuration to the
image.config.openshift.io
custom resource file:spec: registrySources: blockedRegistries: - quay.io/openshift-payload
Verification
Verify that the upstream payload registry is blocked by checking the
/etc/containers/registries.conf
file on the node.Example output
[[registry]] prefix = "" location = "quay.io/openshift-payload" blocked = true mirror-by-digest-only = true [[registry.mirror]] location = "internal-mirror.io/openshift-payload"
9.2.3. Allowing insecure registries
You can add insecure registries, and optionally an individual repository within a registry, by editing the image.config.openshift.io/cluster
custom resource (CR). OpenShift Container Platform applies the changes to this CR to all nodes in the cluster.
Registries that do not use valid SSL certificates or do not require HTTPS connections are considered insecure.
Insecure external registries should be avoided to reduce possible security risks.
Procedure
Edit the
image.config.openshift.io/cluster
custom resource:$ oc edit image.config.openshift.io/cluster
The following is an example
image.config.openshift.io/cluster
CR with an insecure registries list:apiVersion: config.openshift.io/v1 kind: Image metadata: annotations: release.openshift.io/create-only: "true" creationTimestamp: "2019-05-17T13:44:26Z" generation: 1 name: cluster resourceVersion: "8302" selfLink: /apis/config.openshift.io/v1/images/cluster uid: e34555da-78a9-11e9-b92b-06d6c7da38dc spec: registrySources: 1 insecureRegistries: 2 - insecure.com - reg4.io/myrepo/myapp:latest allowedRegistries: - example.com - quay.io - registry.redhat.io - insecure.com 3 - reg4.io/myrepo/myapp:latest - image-registry.openshift-image-registry.svc:5000 status: internalRegistryHostname: image-registry.openshift-image-registry.svc:5000
- 1
- Contains configurations that determine how the container runtime should treat individual registries when accessing images for builds and pods. It does not contain configuration for the internal cluster registry.
- 2
- Specify an insecure registry. You can specify a repository in that registry.
- 3
- Ensure that any insecure registries are included in the
allowedRegistries
list.
NoteWhen the
allowedRegistries
parameter is defined, all registries, including the registry.redhat.io and quay.io registries and the default OpenShift image registry, are blocked unless explicitly listed. If you use the parameter, to prevent pod failure, add all registries including theregistry.redhat.io
andquay.io
registries and theinternalRegistryHostname
to theallowedRegistries
list, as they are required by payload images within your environment. For disconnected clusters, mirror registries should also be added.The Machine Config Operator (MCO) watches the
image.config.openshift.io/cluster
CR for any changes to the registries, then drains and uncordons the nodes when it detects changes. After the nodes return to theReady
state, changes to the insecure and blocked registries appear in the/etc/containers/registries.conf
file on each node.
Verification
To check that the registries have been added to the policy file, use the following command on a node:
$ cat /etc/containers/registries.conf
The following example indicates that images from the
insecure.com
registry is insecure and is allowed for image pulls and pushes.Example output
unqualified-search-registries = ["registry.access.redhat.com", "docker.io"] [[registry]] prefix = "" location = "insecure.com" insecure = true
9.2.4. Adding registries that allow image short names
You can add registries to search for an image short name by editing the image.config.openshift.io/cluster
custom resource (CR). OpenShift Container Platform applies the changes to this CR to all nodes in the cluster.
An image short name enables you to search for images without including the fully qualified domain name in the pull spec. For example, you could use rhel7/etcd
instead of registry.access.redhat.com/rhe7/etcd
.
You might use short names in situations where using the full path is not practical. For example, if your cluster references multiple internal registries whose DNS changes frequently, you would need to update the fully qualified domain names in your pull specs with each change. In this case, using an image short name might be beneficial.
When pulling or pushing images, the container runtime searches the registries listed under the registrySources
parameter in the image.config.openshift.io/cluster
CR. If you created a list of registries under the containerRuntimeSearchRegistries
parameter, when pulling an image with a short name, the container runtime searches those registries.
Using image short names with public registries is strongly discouraged because the image might not deploy if the public registry requires authentication. Use fully-qualified image names with public registries.
Red Hat internal or private registries typically support the use of image short names.
If you list public registries under the containerRuntimeSearchRegistries
parameter (including the registry.redhat.io
, docker.io
, and quay.io
registries), you expose your credentials to all the registries on the list, and you risk network and registry attacks. Because you can only have one pull secret for pulling images, as defined by the global pull secret, that secret is used to authenticate against every registry in that list. Therefore, if you include public registries in the list, you introduce a security risk.
You cannot list multiple public registries under the containerRuntimeSearchRegistries
parameter if each public registry requires different credentials and a cluster does not list the public registry in the global pull secret.
For a public registry that requires authentication, you can use an image short name only if the registry has its credentials stored in the global pull secret.
The Machine Config Operator (MCO) watches the image.config.openshift.io/cluster
resource for any changes to the registries. When the MCO detects a change, it drains the nodes, applies the change, and uncordons the nodes. After the nodes return to the Ready
state, if the containerRuntimeSearchRegistries
parameter is added, the MCO creates a file in the /etc/containers/registries.conf.d
directory on each node with the listed registries. The file overrides the default list of unqualified search registries in the /etc/containers/registries.conf
file. There is no way to fall back to the default list of unqualified search registries.
The containerRuntimeSearchRegistries
parameter works only with the Podman and CRI-O container engines. The registries in the list can be used only in pod specs, not in builds and image streams.
Procedure
Edit the
image.config.openshift.io/cluster
custom resource:$ oc edit image.config.openshift.io/cluster
The following is an example
image.config.openshift.io/cluster
CR:apiVersion: config.openshift.io/v1 kind: Image metadata: annotations: release.openshift.io/create-only: "true" creationTimestamp: "2019-05-17T13:44:26Z" generation: 1 name: cluster resourceVersion: "8302" selfLink: /apis/config.openshift.io/v1/images/cluster uid: e34555da-78a9-11e9-b92b-06d6c7da38dc spec: allowedRegistriesForImport: - domainName: quay.io insecure: false additionalTrustedCA: name: myconfigmap registrySources: containerRuntimeSearchRegistries: 1 - reg1.io - reg2.io - reg3.io allowedRegistries: 2 - example.com - quay.io - registry.redhat.io - reg1.io - reg2.io - reg3.io - image-registry.openshift-image-registry.svc:5000 ... status: internalRegistryHostname: image-registry.openshift-image-registry.svc:5000
NoteWhen the
allowedRegistries
parameter is defined, all registries, including theregistry.redhat.io
andquay.io
registries and the default OpenShift image registry, are blocked unless explicitly listed. If you use this parameter, to prevent pod failure, add all registries including theregistry.redhat.io
andquay.io
registries and theinternalRegistryHostname
to theallowedRegistries
list, as they are required by payload images within your environment. For disconnected clusters, mirror registries should also be added.
Verification
Enter the following command to obtain a list of your nodes:
$ oc get nodes
Example output
NAME STATUS ROLES AGE VERSION <node_name> Ready control-plane,master 37m v1.27.8+4fab27b
Run the following command to enter debug mode on the node:
$ oc debug node/<node_name>
When prompted, enter
chroot /host
into the terminal:sh-4.4# chroot /host
Enter the following command to check that the registries have been added to the policy file:
sh-5.1# cat /etc/containers/registries.conf.d/01-image-searchRegistries.conf
Example output
unqualified-search-registries = ['reg1.io', 'reg2.io', 'reg3.io']
9.2.5. Configuring additional trust stores for image registry access
The image.config.openshift.io/cluster
custom resource can contain a reference to a config map that contains additional certificate authorities to be trusted during image registry access.
Prerequisites
- The certificate authorities (CA) must be PEM-encoded.
Procedure
You can create a config map in the openshift-config
namespace and use its name in AdditionalTrustedCA
in the image.config.openshift.io
custom resource to provide additional CAs that should be trusted when contacting external registries.
The config map key is the hostname of a registry with the port for which this CA is to be trusted, and the PEM certificate content is the value, for each additional registry CA to trust.
Image registry CA config map example
apiVersion: v1
kind: ConfigMap
metadata:
name: my-registry-ca
data:
registry.example.com: |
-----BEGIN CERTIFICATE-----
...
-----END CERTIFICATE-----
registry-with-port.example.com..5000: | 1
-----BEGIN CERTIFICATE-----
...
-----END CERTIFICATE-----
- 1
- If the registry has the port, such as
registry-with-port.example.com:5000
,:
should be replaced with..
.
You can configure additional CAs with the following procedure.
To configure an additional CA:
$ oc create configmap registry-config --from-file=<external_registry_address>=ca.crt -n openshift-config
$ oc edit image.config.openshift.io cluster
spec: additionalTrustedCA: name: registry-config
9.3. Understanding image registry repository mirroring
Setting up container registry repository mirroring enables you to perform the following tasks:
- Configure your OpenShift Container Platform cluster to redirect requests to pull images from a repository on a source image registry and have it resolved by a repository on a mirrored image registry.
- Identify multiple mirrored repositories for each target repository, to make sure that if one mirror is down, another can be used.
Repository mirroring in OpenShift Container Platform includes the following attributes:
- Image pulls are resilient to registry downtimes.
- Clusters in disconnected environments can pull images from critical locations, such as quay.io, and have registries behind a company firewall provide the requested images.
- A particular order of registries is tried when an image pull request is made, with the permanent registry typically being the last one tried.
-
The mirror information you enter is added to the
/etc/containers/registries.conf
file on every node in the OpenShift Container Platform cluster. - When a node makes a request for an image from the source repository, it tries each mirrored repository in turn until it finds the requested content. If all mirrors fail, the cluster tries the source repository. If successful, the image is pulled to the node.
Setting up repository mirroring can be done in the following ways:
At OpenShift Container Platform installation:
By pulling container images needed by OpenShift Container Platform and then bringing those images behind your company’s firewall, you can install OpenShift Container Platform into a datacenter that is in a disconnected environment.
After OpenShift Container Platform installation:
If you did not configure mirroring during OpenShift Container Platform installation, you can do so postinstallation by using any of the following custom resource (CR) objects:
-
ImageDigestMirrorSet
(IDMS). This object allows you to pull images from a mirrored registry by using digest specifications. The IDMS CR enables you to set a fall back policy that allows or stops continued attempts to pull from the source registry if the image pull fails. -
ImageTagMirrorSet
(ITMS). This object allows you to pull images from a mirrored registry by using image tags. The ITMS CR enables you to set a fall back policy that allows or stops continued attempts to pull from the source registry if the image pull fails. -
ImageContentSourcePolicy
(ICSP). This object allows you to pull images from a mirrored registry by using digest specifications. The ICSP CR always falls back to the source registry if the mirrors do not work.
ImportantUsing an
ImageContentSourcePolicy
(ICSP) object to configure repository mirroring is a deprecated feature. Deprecated functionality is still included in OpenShift Container Platform and continues to be supported; however, it will be removed in a future release of this product and is not recommended for new deployments. If you have existing YAML files that you used to createImageContentSourcePolicy
objects, you can use theoc adm migrate icsp
command to convert those files to anImageDigestMirrorSet
YAML file. For more information, see "Converting ImageContentSourcePolicy (ICSP) files for image registry repository mirroring" in the following section.-
Each of these custom resource objects identify the following information:
- The source of the container image repository you want to mirror.
- A separate entry for each mirror repository you want to offer the content requested from the source repository.
For new clusters, you can use IDMS, ITMS, and ICSP CRs objects as desired. However, using IDMS and ITMS is recommended.
If you upgraded a cluster, any existing ICSP objects remain stable, and both IDMS and ICSP objects are supported. Workloads using ICSP objects continue to function as expected. However, if you want to take advantage of the fallback policies introduced in the IDMS CRs, you can migrate current workloads to IDMS objects by using the oc adm migrate icsp
command as shown in the Converting ImageContentSourcePolicy (ICSP) files for image registry repository mirroring section that follows. Migrating to IDMS objects does not require a cluster reboot.
If your cluster uses an ImageDigestMirrorSet
, ImageTagMirrorSet
, or ImageContentSourcePolicy
object to configure repository mirroring, you can use only global pull secrets for mirrored registries. You cannot add a pull secret to a project.
Additional resources
- For more information about global pull secrets, see Updating the global cluster pull secret.
9.3.1. Configuring image registry repository mirroring
You can create postinstallation mirror configuration custom resources (CR) to redirect image pull requests from a source image registry to a mirrored image registry.
Prerequisites
-
Access to the cluster as a user with the
cluster-admin
role.
Procedure
Configure mirrored repositories, by either:
- Setting up a mirrored repository with Red Hat Quay, as described in Red Hat Quay Repository Mirroring. Using Red Hat Quay allows you to copy images from one repository to another and also automatically sync those repositories repeatedly over time.
Using a tool such as
skopeo
to copy images manually from the source repository to the mirrored repository.For example, after installing the skopeo RPM package on a Red Hat Enterprise Linux (RHEL) 7 or RHEL 8 system, use the
skopeo
command as shown in this example:$ skopeo copy \ docker://registry.access.redhat.com/ubi9/ubi-minimal:latest@sha256:5cf... \ docker://example.io/example/ubi-minimal
In this example, you have a container image registry that is named
example.io
with an image repository namedexample
to which you want to copy theubi9/ubi-minimal
image fromregistry.access.redhat.com
. After you create the mirrored registry, you can configure your OpenShift Container Platform cluster to redirect requests made of the source repository to the mirrored repository.
- Log in to your OpenShift Container Platform cluster.
Create a postinstallation mirror configuration CR, by using one of the following examples:
Create an
ImageDigestMirrorSet
orImageTagMirrorSet
CR, as needed, replacing the source and mirrors with your own registry and repository pairs and images:apiVersion: config.openshift.io/v1 1 kind: ImageDigestMirrorSet 2 metadata: name: ubi9repo spec: imageDigestMirrors: 3 - mirrors: - example.io/example/ubi-minimal 4 - example.com/example/ubi-minimal 5 source: registry.access.redhat.com/ubi9/ubi-minimal 6 mirrorSourcePolicy: AllowContactingSource 7 - mirrors: - mirror.example.com/redhat source: registry.example.com/redhat 8 mirrorSourcePolicy: AllowContactingSource - mirrors: - mirror.example.com source: registry.example.com 9 mirrorSourcePolicy: AllowContactingSource - mirrors: - mirror.example.net/image source: registry.example.com/example/myimage 10 mirrorSourcePolicy: AllowContactingSource - mirrors: - mirror.example.net source: registry.example.com/example 11 mirrorSourcePolicy: AllowContactingSource - mirrors: - mirror.example.net/registry-example-com source: registry.example.com 12 mirrorSourcePolicy: AllowContactingSource
- 1
- Indicates the API to use with this CR. This must be
config.openshift.io/v1
. - 2
- Indicates the kind of object according to the pull type:
-
ImageDigestMirrorSet
: Pulls a digest reference image. -
ImageTagMirrorSet
: Pulls a tag reference image.
-
- 3
- Indicates the type of image pull method, either:
-
imageDigestMirrors
: Use for anImageDigestMirrorSet
CR. -
imageTagMirrors
: Use for anImageTagMirrorSet
CR.
-
- 4
- Indicates the name of the mirrored image registry and repository.
- 5
- Optional: Indicates a secondary mirror repository for each target repository. If one mirror is down, the target repository can use another mirror.
- 6
- Indicates the registry and repository source, which is the repository that is referred to in image pull specifications.
- 7
- Optional: Indicates the fallback policy if the image pull fails:
-
AllowContactingSource
: Allows continued attempts to pull the image from the source repository. This is the default. -
NeverContactSource
: Prevents continued attempts to pull the image from the source repository.
-
- 8
- Optional: Indicates a namespace inside a registry, which allows you to use any image in that namespace. If you use a registry domain as a source, the object is applied to all repositories from the registry.
- 9
- Optional: Indicates a registry, which allows you to use any image in that registry. If you specify a registry name, the object is applied to all repositories from a source registry to a mirror registry.
- 10
- Pulls the image
registry.example.com/example/myimage@sha256:…
from the mirrormirror.example.net/image@sha256:..
. - 11
- Pulls the image
registry.example.com/example/image@sha256:…
in the source registry namespace from the mirrormirror.example.net/image@sha256:…
. - 12
- Pulls the image
registry.example.com/myimage@sha256
from the mirror registryexample.net/registry-example-com/myimage@sha256:…
.
Create an
ImageContentSourcePolicy
custom resource, replacing the source and mirrors with your own registry and repository pairs and images:apiVersion: operator.openshift.io/v1alpha1 kind: ImageContentSourcePolicy metadata: name: mirror-ocp spec: repositoryDigestMirrors: - mirrors: - mirror.registry.com:443/ocp/release 1 source: quay.io/openshift-release-dev/ocp-release 2 - mirrors: - mirror.registry.com:443/ocp/release source: quay.io/openshift-release-dev/ocp-v4.0-art-dev
Create the new object:
$ oc create -f registryrepomirror.yaml
After the object is created, the Machine Config Operator (MCO) drains the nodes for
ImageTagMirrorSet
objects only. The MCO does not drain the nodes forImageDigestMirrorSet
andImageContentSourcePolicy
objects.To check that the mirrored configuration settings are applied, do the following on one of the nodes.
List your nodes:
$ oc get node
Example output
NAME STATUS ROLES AGE VERSION ip-10-0-137-44.ec2.internal Ready worker 7m v1.28.5 ip-10-0-138-148.ec2.internal Ready master 11m v1.28.5 ip-10-0-139-122.ec2.internal Ready master 11m v1.28.5 ip-10-0-147-35.ec2.internal Ready worker 7m v1.28.5 ip-10-0-153-12.ec2.internal Ready worker 7m v1.28.5 ip-10-0-154-10.ec2.internal Ready master 11m v1.28.5
Start the debugging process to access the node:
$ oc debug node/ip-10-0-147-35.ec2.internal
Example output
Starting pod/ip-10-0-147-35ec2internal-debug ... To use host binaries, run `chroot /host`
Change your root directory to
/host
:sh-4.2# chroot /host
Check the
/etc/containers/registries.conf
file to make sure the changes were made:sh-4.2# cat /etc/containers/registries.conf
The following output represents a
registries.conf
file where postinstallation mirror configuration CRs were applied. The final two entries are markeddigest-only
andtag-only
respectively.Example output
unqualified-search-registries = ["registry.access.redhat.com", "docker.io"] short-name-mode = "" [[registry]] prefix = "" location = "registry.access.redhat.com/ubi9/ubi-minimal" 1 [[registry.mirror]] location = "example.io/example/ubi-minimal" 2 pull-from-mirror = "digest-only" 3 [[registry.mirror]] location = "example.com/example/ubi-minimal" pull-from-mirror = "digest-only" [[registry]] prefix = "" location = "registry.example.com" [[registry.mirror]] location = "mirror.example.net/registry-example-com" pull-from-mirror = "digest-only" [[registry]] prefix = "" location = "registry.example.com/example" [[registry.mirror]] location = "mirror.example.net" pull-from-mirror = "digest-only" [[registry]] prefix = "" location = "registry.example.com/example/myimage" [[registry.mirror]] location = "mirror.example.net/image" pull-from-mirror = "digest-only" [[registry]] prefix = "" location = "registry.example.com" [[registry.mirror]] location = "mirror.example.com" pull-from-mirror = "digest-only" [[registry]] prefix = "" location = "registry.example.com/redhat" [[registry.mirror]] location = "mirror.example.com/redhat" pull-from-mirror = "digest-only" [[registry]] prefix = "" location = "registry.access.redhat.com/ubi9/ubi-minimal" blocked = true 4 [[registry.mirror]] location = "example.io/example/ubi-minimal-tag" pull-from-mirror = "tag-only" 5
- 1
- Indicates the repository that is referred to in a pull spec.
- 2
- Indicates the mirror for that repository.
- 3
- Indicates that the image pull from the mirror is a digest reference image.
- 4
- Indicates that the
NeverContactSource
parameter is set for this repository. - 5
- Indicates that the image pull from the mirror is a tag reference image.
Pull an image to the node from the source and check if it is resolved by the mirror.
sh-4.2# podman pull --log-level=debug registry.access.redhat.com/ubi9/ubi-minimal@sha256:5cf...
Troubleshooting repository mirroring
If the repository mirroring procedure does not work as described, use the following information about how repository mirroring works to help troubleshoot the problem.
- The first working mirror is used to supply the pulled image.
- The main registry is only used if no other mirror works.
-
From the system context, the
Insecure
flags are used as fallback. -
The format of the
/etc/containers/registries.conf
file has changed recently. It is now version 2 and in TOML format.
9.3.2. Converting ImageContentSourcePolicy (ICSP) files for image registry repository mirroring
Using an ImageContentSourcePolicy
(ICSP) object to configure repository mirroring is a deprecated feature. This functionality is still included in OpenShift Container Platform and continues to be supported; however, it will be removed in a future release of this product and is not recommended for new deployments.
ICSP objects are being replaced by ImageDigestMirrorSet
and ImageTagMirrorSet
objects to configure repository mirroring. If you have existing YAML files that you used to create ImageContentSourcePolicy
objects, you can use the oc adm migrate icsp
command to convert those files to an ImageDigestMirrorSet
YAML file. The command updates the API to the current version, changes the kind
value to ImageDigestMirrorSet
, and changes spec.repositoryDigestMirrors
to spec.imageDigestMirrors
. The rest of the file is not changed.
Because the migration does not change the registries.conf
file, the cluster does not need to reboot.
For more information about ImageDigestMirrorSet
or ImageTagMirrorSet
objects, see "Configuring image registry repository mirroring" in the previous section.
Prerequisites
-
Access to the cluster as a user with the
cluster-admin
role. -
Ensure that you have
ImageContentSourcePolicy
objects on your cluster.
Procedure
Use the following command to convert one or more
ImageContentSourcePolicy
YAML files to anImageDigestMirrorSet
YAML file:$ oc adm migrate icsp <file_name>.yaml <file_name>.yaml <file_name>.yaml --dest-dir <path_to_the_directory>
where:
<file_name>
-
Specifies the name of the source
ImageContentSourcePolicy
YAML. You can list multiple file names. --dest-dir
-
Optional: Specifies a directory for the output
ImageDigestMirrorSet
YAML. If unset, the file is written to the current directory.
For example, the following command converts the
icsp.yaml
andicsp-2.yaml
file and saves the new YAML files to theidms-files
directory.$ oc adm migrate icsp icsp.yaml icsp-2.yaml --dest-dir idms-files
Example output
wrote ImageDigestMirrorSet to idms-files/imagedigestmirrorset_ubi8repo.5911620242173376087.yaml wrote ImageDigestMirrorSet to idms-files/imagedigestmirrorset_ubi9repo.6456931852378115011.yaml
Create the CR object by running the following command:
$ oc create -f <path_to_the_directory>/<file-name>.yaml
where:
<path_to_the_directory>
-
Specifies the path to the directory, if you used the
--dest-dir
flag. <file_name>
-
Specifies the name of the
ImageDigestMirrorSet
YAML.
- Remove the ICSP objects after the IDMS objects are rolled out.
Chapter 10. Using templates
The following sections provide an overview of templates, as well as how to use and create them.
10.1. Understanding templates
A template describes a set of objects that can be parameterized and processed to produce a list of objects for creation by OpenShift Container Platform. A template can be processed to create anything you have permission to create within a project, for example services, build configurations, and deployment configurations. A template can also define a set of labels to apply to every object defined in the template.
You can create a list of objects from a template using the CLI or, if a template has been uploaded to your project or the global template library, using the web console.
10.2. Uploading a template
If you have a JSON or YAML file that defines a template, you can upload the template to projects using the CLI. This saves the template to the project for repeated use by any user with appropriate access to that project. Instructions about writing your own templates are provided later in this topic.
Procedure
Upload a template using one of the following methods:
Upload a template to your current project’s template library, pass the JSON or YAML file with the following command:
$ oc create -f <filename>
Upload a template to a different project using the
-n
option with the name of the project:$ oc create -f <filename> -n <project>
The template is now available for selection using the web console or the CLI.
10.3. Creating an application by using the web console
You can use the web console to create an application from a template.
Procedure
- Select Developer from the context selector at the top of the web console navigation menu.
- While in the desired project, click +Add
- Click All services in the Developer Catalog tile.
Click Builder Images under Type to see the available builder images.
NoteOnly image stream tags that have the
builder
tag listed in their annotations appear in this list, as demonstrated here:kind: "ImageStream" apiVersion: "v1" metadata: name: "ruby" creationTimestamp: null spec: # ... tags: - name: "2.6" annotations: description: "Build and run Ruby 2.6 applications" iconClass: "icon-ruby" tags: "builder,ruby" 1 supports: "ruby:2.6,ruby" version: "2.6" # ...
- 1
- Including
builder
here ensures this image stream tag appears in the web console as a builder.
- Modify the settings in the new application screen to configure the objects to support your application.
10.4. Creating objects from templates by using the CLI
You can use the CLI to process templates and use the configuration that is generated to create objects.
10.4.1. Adding labels
Labels are used to manage and organize generated objects, such as pods. The labels specified in the template are applied to every object that is generated from the template.
Procedure
Add labels in the template from the command line:
$ oc process -f <filename> -l name=otherLabel
10.4.2. Listing parameters
The list of parameters that you can override are listed in the parameters
section of the template.
Procedure
You can list parameters with the CLI by using the following command and specifying the file to be used:
$ oc process --parameters -f <filename>
Alternatively, if the template is already uploaded:
$ oc process --parameters -n <project> <template_name>
For example, the following shows the output when listing the parameters for one of the quick start templates in the default
openshift
project:$ oc process --parameters -n openshift rails-postgresql-example
Example output
NAME DESCRIPTION GENERATOR VALUE SOURCE_REPOSITORY_URL The URL of the repository with your application source code https://github.com/sclorg/rails-ex.git SOURCE_REPOSITORY_REF Set this to a branch name, tag or other ref of your repository if you are not using the default branch CONTEXT_DIR Set this to the relative path to your project if it is not in the root of your repository APPLICATION_DOMAIN The exposed hostname that will route to the Rails service rails-postgresql-example.openshiftapps.com GITHUB_WEBHOOK_SECRET A secret string used to configure the GitHub webhook expression [a-zA-Z0-9]{40} SECRET_KEY_BASE Your secret key for verifying the integrity of signed cookies expression [a-z0-9]{127} APPLICATION_USER The application user that is used within the sample application to authorize access on pages openshift APPLICATION_PASSWORD The application password that is used within the sample application to authorize access on pages secret DATABASE_SERVICE_NAME Database service name postgresql POSTGRESQL_USER database username expression user[A-Z0-9]{3} POSTGRESQL_PASSWORD database password expression [a-zA-Z0-9]{8} POSTGRESQL_DATABASE database name root POSTGRESQL_MAX_CONNECTIONS database max connections 10 POSTGRESQL_SHARED_BUFFERS database shared buffers 12MB
The output identifies several parameters that are generated with a regular expression-like generator when the template is processed.
10.4.3. Generating a list of objects
Using the CLI, you can process a file defining a template to return the list of objects to standard output.
Procedure
Process a file defining a template to return the list of objects to standard output:
$ oc process -f <filename>
Alternatively, if the template has already been uploaded to the current project:
$ oc process <template_name>
Create objects from a template by processing the template and piping the output to
oc create
:$ oc process -f <filename> | oc create -f -
Alternatively, if the template has already been uploaded to the current project:
$ oc process <template> | oc create -f -
You can override any parameter values defined in the file by adding the
-p
option for each<name>=<value>
pair you want to override. A parameter reference appears in any text field inside the template items.For example, in the following the
POSTGRESQL_USER
andPOSTGRESQL_DATABASE
parameters of a template are overridden to output a configuration with customized environment variables:Creating a List of objects from a template
$ oc process -f my-rails-postgresql \ -p POSTGRESQL_USER=bob \ -p POSTGRESQL_DATABASE=mydatabase
The JSON file can either be redirected to a file or applied directly without uploading the template by piping the processed output to the
oc create
command:$ oc process -f my-rails-postgresql \ -p POSTGRESQL_USER=bob \ -p POSTGRESQL_DATABASE=mydatabase \ | oc create -f -
If you have large number of parameters, you can store them in a file and then pass this file to
oc process
:$ cat postgres.env POSTGRESQL_USER=bob POSTGRESQL_DATABASE=mydatabase
$ oc process -f my-rails-postgresql --param-file=postgres.env
You can also read the environment from standard input by using
"-"
as the argument to--param-file
:$ sed s/bob/alice/ postgres.env | oc process -f my-rails-postgresql --param-file=-
10.5. Modifying uploaded templates
You can edit a template that has already been uploaded to your project.
Procedure
Modify a template that has already been uploaded:
$ oc edit template <template>
10.6. Using instant app and quick start templates
OpenShift Container Platform provides a number of default instant app and quick start templates to make it easy to quickly get started creating a new application for different languages. Templates are provided for Rails (Ruby), Django (Python), Node.js, CakePHP (PHP), and Dancer (Perl). Your cluster administrator must create these templates in the default, global openshift
project so you have access to them.
By default, the templates build using a public source repository on GitHub that contains the necessary application code.
Procedure
You can list the available default instant app and quick start templates with:
$ oc get templates -n openshift
To modify the source and build your own version of the application:
-
Fork the repository referenced by the template’s default
SOURCE_REPOSITORY_URL
parameter. Override the value of the
SOURCE_REPOSITORY_URL
parameter when creating from the template, specifying your fork instead of the default value.By doing this, the build configuration created by the template now points to your fork of the application code, and you can modify the code and rebuild the application at will.
-
Fork the repository referenced by the template’s default
Some of the instant app and quick start templates define a database deployment configuration. The configuration they define uses ephemeral storage for the database content. These templates should be used for demonstration purposes only as all database data is lost if the database pod restarts for any reason.
10.6.1. Quick start templates
A quick start template is a basic example of an application running on OpenShift Container Platform. Quick starts come in a variety of languages and frameworks, and are defined in a template, which is constructed from a set of services, build configurations, and deployment configurations. This template references the necessary images and source repositories to build and deploy the application.
To explore a quick start, create an application from a template. Your administrator must have already installed these templates in your OpenShift Container Platform cluster, in which case you can simply select it from the web console.
Quick starts refer to a source repository that contains the application source code. To customize the quick start, fork the repository and, when creating an application from the template, substitute the default source repository name with your forked repository. This results in builds that are performed using your source code instead of the provided example source. You can then update the code in your source repository and launch a new build to see the changes reflected in the deployed application.
10.6.1.1. Web framework quick start templates
These quick start templates provide a basic application of the indicated framework and language:
- CakePHP: a PHP web framework that includes a MySQL database
- Dancer: a Perl web framework that includes a MySQL database
- Django: a Python web framework that includes a PostgreSQL database
- NodeJS: a NodeJS web application that includes a MongoDB database
- Rails: a Ruby web framework that includes a PostgreSQL database
10.7. Writing templates
You can define new templates to make it easy to recreate all the objects of your application. The template defines the objects it creates along with some metadata to guide the creation of those objects.
The following is an example of a simple template object definition (YAML):
apiVersion: template.openshift.io/v1 kind: Template metadata: name: redis-template annotations: description: "Description" iconClass: "icon-redis" tags: "database,nosql" objects: - apiVersion: v1 kind: Pod metadata: name: redis-master spec: containers: - env: - name: REDIS_PASSWORD value: ${REDIS_PASSWORD} image: dockerfile/redis name: master ports: - containerPort: 6379 protocol: TCP parameters: - description: Password used for Redis authentication from: '[A-Z0-9]{8}' generate: expression name: REDIS_PASSWORD labels: redis: master
10.7.1. Writing the template description
The template description informs you what the template does and helps you find it when searching in the web console. Additional metadata beyond the template name is optional, but useful to have. In addition to general descriptive information, the metadata also includes a set of tags. Useful tags include the name of the language the template is related to for example, Java, PHP, Ruby, and so on.
The following is an example of template description metadata:
kind: Template apiVersion: template.openshift.io/v1 metadata: name: cakephp-mysql-example 1 annotations: openshift.io/display-name: "CakePHP MySQL Example (Ephemeral)" 2 description: >- An example CakePHP application with a MySQL database. For more information about using this template, including OpenShift considerations, see https://github.com/sclorg/cakephp-ex/blob/master/README.md. WARNING: Any data stored will be lost upon pod destruction. Only use this template for testing." 3 openshift.io/long-description: >- This template defines resources needed to develop a CakePHP application, including a build configuration, application DeploymentConfig, and database DeploymentConfig. The database is stored in non-persistent storage, so this configuration should be used for experimental purposes only. 4 tags: "quickstart,php,cakephp" 5 iconClass: icon-php 6 openshift.io/provider-display-name: "Red Hat, Inc." 7 openshift.io/documentation-url: "https://github.com/sclorg/cakephp-ex" 8 openshift.io/support-url: "https://access.redhat.com" 9 message: "Your admin credentials are ${ADMIN_USERNAME}:${ADMIN_PASSWORD}" 10
- 1
- The unique name of the template.
- 2
- A brief, user-friendly name, which can be employed by user interfaces.
- 3
- A description of the template. Include enough detail that users understand what is being deployed and any caveats they must know before deploying. It should also provide links to additional information, such as a README file. Newlines can be included to create paragraphs.
- 4
- Additional template description. This may be displayed by the service catalog, for example.
- 5
- Tags to be associated with the template for searching and grouping. Add tags that include it into one of the provided catalog categories. Refer to the
id
andcategoryAliases
inCATALOG_CATEGORIES
in the console constants file. The categories can also be customized for the whole cluster. - 6
- An icon to be displayed with your template in the web console.
Example 10.1. Available icons
-
icon-3scale
-
icon-aerogear
-
icon-amq
-
icon-angularjs
-
icon-ansible
-
icon-apache
-
icon-beaker
-
icon-camel
-
icon-capedwarf
-
icon-cassandra
-
icon-catalog-icon
-
icon-clojure
-
icon-codeigniter
-
icon-cordova
-
icon-datagrid
-
icon-datavirt
-
icon-debian
-
icon-decisionserver
-
icon-django
-
icon-dotnet
-
icon-drupal
-
icon-eap
-
icon-elastic
-
icon-erlang
-
icon-fedora
-
icon-freebsd
-
icon-git
-
icon-github
-
icon-gitlab
-
icon-glassfish
-
icon-go-gopher
-
icon-golang
-
icon-grails
-
icon-hadoop
-
icon-haproxy
-
icon-helm
-
icon-infinispan
-
icon-jboss
-
icon-jenkins
-
icon-jetty
-
icon-joomla
-
icon-jruby
-
icon-js
-
icon-knative
-
icon-kubevirt
-
icon-laravel
-
icon-load-balancer
-
icon-mariadb
-
icon-mediawiki
-
icon-memcached
-
icon-mongodb
-
icon-mssql
-
icon-mysql-database
-
icon-nginx
-
icon-nodejs
-
icon-openjdk
-
icon-openliberty
-
icon-openshift
-
icon-openstack
-
icon-other-linux
-
icon-other-unknown
-
icon-perl
-
icon-phalcon
-
icon-php
-
icon-play
-
iconpostgresql
-
icon-processserver
-
icon-python
-
icon-quarkus
-
icon-rabbitmq
-
icon-rails
-
icon-redhat
-
icon-redis
-
icon-rh-integration
-
icon-rh-spring-boot
-
icon-rh-tomcat
-
icon-ruby
-
icon-scala
-
icon-serverlessfx
-
icon-shadowman
-
icon-spring-boot
-
icon-spring
-
icon-sso
-
icon-stackoverflow
-
icon-suse
-
icon-symfony
-
icon-tomcat
-
icon-ubuntu
-
icon-vertx
-
icon-wildfly
-
icon-windows
-
icon-wordpress
-
icon-xamarin
-
icon-zend
-
- 7
- The name of the person or organization providing the template.
- 8
- A URL referencing further documentation for the template.
- 9
- A URL where support can be obtained for the template.
- 10
- An instructional message that is displayed when this template is instantiated. This field should inform the user how to use the newly created resources. Parameter substitution is performed on the message before being displayed so that generated credentials and other parameters can be included in the output. Include links to any next-steps documentation that users should follow.
10.7.2. Writing template labels
Templates can include a set of labels. These labels are added to each object created when the template is instantiated. Defining a label in this way makes it easy for users to find and manage all the objects created from a particular template.
The following is an example of template object labels:
kind: "Template" apiVersion: "v1" ... labels: template: "cakephp-mysql-example" 1 app: "${NAME}" 2
10.7.3. Writing template parameters
Parameters allow a value to be supplied by you or generated when the template is instantiated. Then, that value is substituted wherever the parameter is referenced. References can be defined in any field in the objects list field. This is useful for generating random passwords or allowing you to supply a hostname or other user-specific value that is required to customize the template. Parameters can be referenced in two ways:
-
As a string value by placing values in the form
${PARAMETER_NAME}
in any string field in the template. -
As a JSON or YAML value by placing values in the form
${{PARAMETER_NAME}}
in place of any field in the template.
When using the ${PARAMETER_NAME}
syntax, multiple parameter references can be combined in a single field and the reference can be embedded within fixed data, such as "http://${PARAMETER_1}${PARAMETER_2}"
. Both parameter values are substituted and the resulting value is a quoted string.
When using the ${{PARAMETER_NAME}}
syntax only a single parameter reference is allowed and leading and trailing characters are not permitted. The resulting value is unquoted unless, after substitution is performed, the result is not a valid JSON object. If the result is not a valid JSON value, the resulting value is quoted and treated as a standard string.
A single parameter can be referenced multiple times within a template and it can be referenced using both substitution syntaxes within a single template.
A default value can be provided, which is used if you do not supply a different value:
The following is an example of setting an explicit value as the default value:
parameters: - name: USERNAME description: "The user name for Joe" value: joe
Parameter values can also be generated based on rules specified in the parameter definition, for example generating a parameter value:
parameters: - name: PASSWORD description: "The random user password" generate: expression from: "[a-zA-Z0-9]{12}"
In the previous example, processing generates a random password 12 characters long consisting of all upper and lowercase alphabet letters and numbers.
The syntax available is not a full regular expression syntax. However, you can use \w
, \d
, \a
, and \A
modifiers:
-
[\w]{10}
produces 10 alphabet characters, numbers, and underscores. This follows the PCRE standard and is equal to[a-zA-Z0-9_]{10}
. -
[\d]{10}
produces 10 numbers. This is equal to[0-9]{10}
. -
[\a]{10}
produces 10 alphabetical characters. This is equal to[a-zA-Z]{10}
. -
[\A]{10}
produces 10 punctuation or symbol characters. This is equal to[~!@#$%\^&*()\-_+={}\[\]\\|<,>.?/"';:`]{10}
.
Depending on if the template is written in YAML or JSON, and the type of string that the modifier is embedded within, you might need to escape the backslash with a second backslash. The following examples are equivalent:
Example YAML template with a modifier
parameters: - name: singlequoted_example generate: expression from: '[\A]{10}' - name: doublequoted_example generate: expression from: "[\\A]{10}"
Example JSON template with a modifier
{ "parameters": [ { "name": "json_example", "generate": "expression", "from": "[\\A]{10}" } ] }
Here is an example of a full template with parameter definitions and references:
kind: Template apiVersion: template.openshift.io/v1 metadata: name: my-template objects: - kind: BuildConfig apiVersion: build.openshift.io/v1 metadata: name: cakephp-mysql-example annotations: description: Defines how to build the application spec: source: type: Git git: uri: "${SOURCE_REPOSITORY_URL}" 1 ref: "${SOURCE_REPOSITORY_REF}" contextDir: "${CONTEXT_DIR}" - kind: DeploymentConfig apiVersion: apps.openshift.io/v1 metadata: name: frontend spec: replicas: "${{REPLICA_COUNT}}" 2 parameters: - name: SOURCE_REPOSITORY_URL 3 displayName: Source Repository URL 4 description: The URL of the repository with your application source code 5 value: https://github.com/sclorg/cakephp-ex.git 6 required: true 7 - name: GITHUB_WEBHOOK_SECRET description: A secret string used to configure the GitHub webhook generate: expression 8 from: "[a-zA-Z0-9]{40}" 9 - name: REPLICA_COUNT description: Number of replicas to run value: "2" required: true message: "... The GitHub webhook secret is ${GITHUB_WEBHOOK_SECRET} ..." 10
- 1
- This value is replaced with the value of the
SOURCE_REPOSITORY_URL
parameter when the template is instantiated. - 2
- This value is replaced with the unquoted value of the
REPLICA_COUNT
parameter when the template is instantiated. - 3
- The name of the parameter. This value is used to reference the parameter within the template.
- 4
- The user-friendly name for the parameter. This is displayed to users.
- 5
- A description of the parameter. Provide more detailed information for the purpose of the parameter, including any constraints on the expected value. Descriptions should use complete sentences to follow the console’s text standards. Do not make this a duplicate of the display name.
- 6
- A default value for the parameter which is used if you do not override the value when instantiating the template. Avoid using default values for things like passwords, instead use generated parameters in combination with secrets.
- 7
- Indicates this parameter is required, meaning you cannot override it with an empty value. If the parameter does not provide a default or generated value, you must supply a value.
- 8
- A parameter which has its value generated.
- 9
- The input to the generator. In this case, the generator produces a 40 character alphanumeric value including upper and lowercase characters.
- 10
- Parameters can be included in the template message. This informs you about generated values.
10.7.4. Writing the template object list
The main portion of the template is the list of objects which is created when the template is instantiated. This can be any valid API object, such as a build configuration, deployment configuration, or service. The object is created exactly as defined here, with any parameter values substituted in prior to creation. The definition of these objects can reference parameters defined earlier.
The following is an example of an object list:
kind: "Template"
apiVersion: "v1"
metadata:
name: my-template
objects:
- kind: "Service" 1
apiVersion: "v1"
metadata:
name: "cakephp-mysql-example"
annotations:
description: "Exposes and load balances the application pods"
spec:
ports:
- name: "web"
port: 8080
targetPort: 8080
selector:
name: "cakephp-mysql-example"
- 1
- The definition of a service, which is created by this template.
If an object definition metadata includes a fixed namespace
field value, the field is stripped out of the definition during template instantiation. If the namespace
field contains a parameter reference, normal parameter substitution is performed and the object is created in whatever namespace the parameter substitution resolved the value to, assuming the user has permission to create objects in that namespace.
10.7.5. Marking a template as bindable
The Template Service Broker advertises one service in its catalog for each template object of which it is aware. By default, each of these services is advertised as being bindable, meaning an end user is permitted to bind against the provisioned service.
Procedure
Template authors can prevent end users from binding against services provisioned from a given template.
-
Prevent end user from binding against services provisioned from a given template by adding the annotation
template.openshift.io/bindable: "false"
to the template.
10.7.6. Exposing template object fields
Template authors can indicate that fields of particular objects in a template should be exposed. The Template Service Broker recognizes exposed fields on ConfigMap
, Secret
, Service
, and Route
objects, and returns the values of the exposed fields when a user binds a service backed by the broker.
To expose one or more fields of an object, add annotations prefixed by template.openshift.io/expose-
or template.openshift.io/base64-expose-
to the object in the template.
Each annotation key, with its prefix removed, is passed through to become a key in a bind
response.
Each annotation value is a Kubernetes JSONPath expression, which is resolved at bind time to indicate the object field whose value should be returned in the bind
response.
Bind
response key-value pairs can be used in other parts of the system as environment variables. Therefore, it is recommended that every annotation key with its prefix removed should be a valid environment variable name — beginning with a character A-Z
, a-z
, or _
, and being followed by zero or more characters A-Z
, a-z
, 0-9
, or _
.
Unless escaped with a backslash, Kubernetes' JSONPath implementation interprets characters such as .
, @
, and others as metacharacters, regardless of their position in the expression. Therefore, for example, to refer to a ConfigMap
datum named my.key
, the required JSONPath expression would be {.data['my\.key']}
. Depending on how the JSONPath expression is then written in YAML, an additional backslash might be required, for example "{.data['my\\.key']}"
.
The following is an example of different objects' fields being exposed:
kind: Template apiVersion: template.openshift.io/v1 metadata: name: my-template objects: - kind: ConfigMap apiVersion: v1 metadata: name: my-template-config annotations: template.openshift.io/expose-username: "{.data['my\\.username']}" data: my.username: foo - kind: Secret apiVersion: v1 metadata: name: my-template-config-secret annotations: template.openshift.io/base64-expose-password: "{.data['password']}" stringData: password: <password> - kind: Service apiVersion: v1 metadata: name: my-template-service annotations: template.openshift.io/expose-service_ip_port: "{.spec.clusterIP}:{.spec.ports[?(.name==\"web\")].port}" spec: ports: - name: "web" port: 8080 - kind: Route apiVersion: route.openshift.io/v1 metadata: name: my-template-route annotations: template.openshift.io/expose-uri: "http://{.spec.host}{.spec.path}" spec: path: mypath
An example response to a bind
operation given the above partial template follows:
{ "credentials": { "username": "foo", "password": "YmFy", "service_ip_port": "172.30.12.34:8080", "uri": "http://route-test.router.default.svc.cluster.local/mypath" } }
Procedure
-
Use the
template.openshift.io/expose-
annotation to return the field value as a string. This is convenient, although it does not handle arbitrary binary data. -
If you want to return binary data, use the
template.openshift.io/base64-expose-
annotation instead to base64 encode the data before it is returned.
10.7.7. Waiting for template readiness
Template authors can indicate that certain objects within a template should be waited for before a template instantiation by the service catalog, Template Service Broker, or TemplateInstance
API is considered complete.
To use this feature, mark one or more objects of kind Build
, BuildConfig
, Deployment
, DeploymentConfig
, Job
, or StatefulSet
in a template with the following annotation:
"template.alpha.openshift.io/wait-for-ready": "true"
Template instantiation is not complete until all objects marked with the annotation report ready. Similarly, if any of the annotated objects report failed, or if the template fails to become ready within a fixed timeout of one hour, the template instantiation fails.
For the purposes of instantiation, readiness and failure of each object kind are defined as follows:
Kind | Readiness | Failure |
---|---|---|
| Object reports phase complete. | Object reports phase canceled, error, or failed. |
| Latest associated build object reports phase complete. | Latest associated build object reports phase canceled, error, or failed. |
| Object reports new replica set and deployment available. This honors readiness probes defined on the object. | Object reports progressing condition as false. |
| Object reports new replication controller and deployment available. This honors readiness probes defined on the object. | Object reports progressing condition as false. |
| Object reports completion. | Object reports that one or more failures have occurred. |
| Object reports all replicas ready. This honors readiness probes defined on the object. | Not applicable. |
The following is an example template extract, which uses the wait-for-ready
annotation. Further examples can be found in the OpenShift Container Platform quick start templates.
kind: Template apiVersion: template.openshift.io/v1 metadata: name: my-template objects: - kind: BuildConfig apiVersion: build.openshift.io/v1 metadata: name: ... annotations: # wait-for-ready used on BuildConfig ensures that template instantiation # will fail immediately if build fails template.alpha.openshift.io/wait-for-ready: "true" spec: ... - kind: DeploymentConfig apiVersion: apps.openshift.io/v1 metadata: name: ... annotations: template.alpha.openshift.io/wait-for-ready: "true" spec: ... - kind: Service apiVersion: v1 metadata: name: ... spec: ...
Additional recommendations
- Set memory, CPU, and storage default sizes to make sure your application is given enough resources to run smoothly.
-
Avoid referencing the
latest
tag from images if that tag is used across major versions. This can cause running applications to break when new images are pushed to that tag. - A good template builds and deploys cleanly without requiring modifications after the template is deployed.
10.7.8. Creating a template from existing objects
Rather than writing an entire template from scratch, you can export existing objects from your project in YAML form, and then modify the YAML from there by adding parameters and other customizations as template form.
Procedure
Export objects in a project in YAML form:
$ oc get -o yaml all > <yaml_filename>
You can also substitute a particular resource type or multiple resources instead of
all
. Runoc get -h
for more examples.The object types included in
oc get -o yaml all
are:-
BuildConfig
-
Build
-
DeploymentConfig
-
ImageStream
-
Pod
-
ReplicationController
-
Route
-
Service
-
Using the all
alias is not recommended because the contents might vary across different clusters and versions. Instead, specify all required resources.
Chapter 11. Using Ruby on Rails
Ruby on Rails is a web framework written in Ruby. This guide covers using Rails 4 on OpenShift Container Platform.
Go through the whole tutorial to have an overview of all the steps necessary to run your application on the OpenShift Container Platform. If you experience a problem try reading through the entire tutorial and then going back to your issue. It can also be useful to review your previous steps to ensure that all the steps were run correctly.
11.1. Prerequisites
- Basic Ruby and Rails knowledge.
- Locally installed version of Ruby 2.0.0+, Rubygems, Bundler.
- Basic Git knowledge.
- Running instance of OpenShift Container Platform 4.
-
Make sure that an instance of OpenShift Container Platform is running and is available. Also make sure that your
oc
CLI client is installed and the command is accessible from your command shell, so you can use it to log in using your email address and password.
11.2. Setting up the database
Rails applications are almost always used with a database. For local development use the PostgreSQL database.
Procedure
Install the database:
$ sudo yum install -y postgresql postgresql-server postgresql-devel
Initialize the database:
$ sudo postgresql-setup initdb
This command creates the
/var/lib/pgsql/data
directory, in which the data is stored.Start the database:
$ sudo systemctl start postgresql.service
When the database is running, create your
rails
user:$ sudo -u postgres createuser -s rails
Note that the user created has no password.
11.3. Writing your application
If you are starting your Rails application from scratch, you must install the Rails gem first. Then you can proceed with writing your application.
Procedure
Install the Rails gem:
$ gem install rails
Example output
Successfully installed rails-4.3.0 1 gem installed
After you install the Rails gem, create a new application with PostgreSQL as your database:
$ rails new rails-app --database=postgresql
Change into your new application directory:
$ cd rails-app
If you already have an application, make sure the
pg
(postgresql) gem is present in yourGemfile
. If not, edit yourGemfile
by adding the gem:gem 'pg'
Generate a new
Gemfile.lock
with all your dependencies:$ bundle install
In addition to using the
postgresql
database with thepg
gem, you also must ensure that theconfig/database.yml
is using thepostgresql
adapter.Make sure you updated
default
section in theconfig/database.yml
file, so it looks like this:default: &default adapter: postgresql encoding: unicode pool: 5 host: localhost username: rails password: <password>
Create your application’s development and test databases:
$ rake db:create
This creates
development
andtest
database in your PostgreSQL server.
11.3.1. Creating a welcome page
Since Rails 4 no longer serves a static public/index.html
page in production, you must create a new root page.
To have a custom welcome page must do following steps:
- Create a controller with an index action.
- Create a view page for the welcome controller index action.
- Create a route that serves applications root page with the created controller and view.
Rails offers a generator that completes all necessary steps for you.
Procedure
Run Rails generator:
$ rails generate controller welcome index
All the necessary files are created.
edit line 2 in
config/routes.rb
file as follows:root 'welcome#index'
Run the rails server to verify the page is available:
$ rails server
You should see your page by visiting http://localhost:3000 in your browser. If you do not see the page, check the logs that are output to your server to debug.
11.3.2. Configuring application for OpenShift Container Platform
To have your application communicate with the PostgreSQL database service running in OpenShift Container Platform you must edit the default
section in your config/database.yml
to use environment variables, which you must define later, upon the database service creation.
Procedure
Edit the
default
section in yourconfig/database.yml
with pre-defined variables as follows:Sample
config/database
YAML file<% user = ENV.key?("POSTGRESQL_ADMIN_PASSWORD") ? "root" : ENV["POSTGRESQL_USER"] %> <% password = ENV.key?("POSTGRESQL_ADMIN_PASSWORD") ? ENV["POSTGRESQL_ADMIN_PASSWORD"] : ENV["POSTGRESQL_PASSWORD"] %> <% db_service = ENV.fetch("DATABASE_SERVICE_NAME","").upcase %> default: &default adapter: postgresql encoding: unicode # For details on connection pooling, see rails configuration guide # http://guides.rubyonrails.org/configuring.html#database-pooling pool: <%= ENV["POSTGRESQL_MAX_CONNECTIONS"] || 5 %> username: <%= user %> password: <%= password %> host: <%= ENV["#{db_service}_SERVICE_HOST"] %> port: <%= ENV["#{db_service}_SERVICE_PORT"] %> database: <%= ENV["POSTGRESQL_DATABASE"] %>
11.3.3. Storing your application in Git
Building an application in OpenShift Container Platform usually requires that the source code be stored in a git repository, so you must install git
if you do not already have it.
Prerequisites
- Install git.
Procedure
Make sure you are in your Rails application directory by running the
ls -1
command. The output of the command should look like:$ ls -1
Example output
app bin config config.ru db Gemfile Gemfile.lock lib log public Rakefile README.rdoc test tmp vendor
Run the following commands in your Rails app directory to initialize and commit your code to git:
$ git init
$ git add .
$ git commit -m "initial commit"
After your application is committed you must push it to a remote repository. GitHub account, in which you create a new repository.
Set the remote that points to your
git
repository:$ git remote add origin git@github.com:<namespace/repository-name>.git
Push your application to your remote git repository.
$ git push
11.4. Deploying your application to OpenShift Container Platform
You can deploy you application to OpenShift Container Platform.
After creating the rails-app
project, you are automatically switched to the new project namespace.
Deploying your application in OpenShift Container Platform involves three steps:
- Creating a database service from OpenShift Container Platform’s PostgreSQL image.
- Creating a frontend service from OpenShift Container Platform’s Ruby 2.0 builder image and your Ruby on Rails source code, which are wired with the database service.
- Creating a route for your application.
Procedure
To deploy your Ruby on Rails application, create a new project for the application:
$ oc new-project rails-app --description="My Rails application" --display-name="Rails Application"
11.4.1. Creating the database service
Your Rails application expects a running database service. For this service use PostgreSQL database image.
To create the database service, use the oc new-app
command. To this command you must pass some necessary environment variables which are used inside the database container. These environment variables are required to set the username, password, and name of the database. You can change the values of these environment variables to anything you would like. The variables are as follows:
- POSTGRESQL_DATABASE
- POSTGRESQL_USER
- POSTGRESQL_PASSWORD
Setting these variables ensures:
- A database exists with the specified name.
- A user exists with the specified name.
- The user can access the specified database with the specified password.
Procedure
Create the database service:
$ oc new-app postgresql -e POSTGRESQL_DATABASE=db_name -e POSTGRESQL_USER=username -e POSTGRESQL_PASSWORD=password
To also set the password for the database administrator, append to the previous command with:
-e POSTGRESQL_ADMIN_PASSWORD=admin_pw
Watch the progress:
$ oc get pods --watch
11.4.2. Creating the frontend service
To bring your application to OpenShift Container Platform, you must specify a repository in which your application lives.
Procedure
Create the frontend service and specify database related environment variables that were setup when creating the database service:
$ oc new-app path/to/source/code --name=rails-app -e POSTGRESQL_USER=username -e POSTGRESQL_PASSWORD=password -e POSTGRESQL_DATABASE=db_name -e DATABASE_SERVICE_NAME=postgresql
With this command, OpenShift Container Platform fetches the source code, sets up the builder, builds your application image, and deploys the newly created image together with the specified environment variables. The application is named
rails-app
.Verify the environment variables have been added by viewing the JSON document of the
rails-app
deployment config:$ oc get dc rails-app -o json
You should see the following section:
Example output
env": [ { "name": "POSTGRESQL_USER", "value": "username" }, { "name": "POSTGRESQL_PASSWORD", "value": "password" }, { "name": "POSTGRESQL_DATABASE", "value": "db_name" }, { "name": "DATABASE_SERVICE_NAME", "value": "postgresql" } ],
Check the build process:
$ oc logs -f build/rails-app-1
After the build is complete, look at the running pods in OpenShift Container Platform:
$ oc get pods
You should see a line starting with
myapp-<number>-<hash>
, and that is your application running in OpenShift Container Platform.Before your application is functional, you must initialize the database by running the database migration script. There are two ways you can do this:
Manually from the running frontend container:
Exec into frontend container with
rsh
command:$ oc rsh <frontend_pod_id>
Run the migration from inside the container:
$ RAILS_ENV=production bundle exec rake db:migrate
If you are running your Rails application in a
development
ortest
environment you do not have to specify theRAILS_ENV
environment variable.
- By adding pre-deployment lifecycle hooks in your template.
11.4.3. Creating a route for your application
You can expose a service to create a route for your application.
Procedure
To expose a service by giving it an externally-reachable hostname like
www.example.com
use OpenShift Container Platform route. In your case you need to expose the frontend service by typing:$ oc expose service rails-app --hostname=www.example.com
Ensure the hostname you specify resolves into the IP address of the router.
Chapter 12. Using images
12.1. Using images overview
Use the following topics to discover the different Source-to-Image (S2I), database, and other container images that are available for OpenShift Container Platform users.
Red Hat official container images are provided in the Red Hat Registry at registry.redhat.io. OpenShift Container Platform’s supported S2I, database, and Jenkins images are provided in the openshift4
repository in the Red Hat Quay Registry. For example, quay.io/openshift-release-dev/ocp-v4.0-<address>
is the name of the OpenShift Application Platform image.
The xPaaS middleware images are provided in their respective product repositories on the Red Hat Registry but suffixed with a -openshift
. For example, registry.redhat.io/jboss-eap-6/eap64-openshift
is the name of the JBoss EAP image.
All Red Hat supported images covered in this section are described in the Container images section of the Red Hat Ecosystem Catalog. For every version of each image, you can find details on its contents and usage. Browse or search for the image that interests you.
The newer versions of container images are not compatible with earlier versions of OpenShift Container Platform. Verify and use the correct version of container images, based on your version of OpenShift Container Platform.
12.2. Source-to-image
You can use the Red Hat Software Collections images as a foundation for applications that rely on specific runtime environments such as Node.js, Perl, or Python. You can use the Red Hat Java Source-to-Image for OpenShift documentation as a reference for runtime environments that use Java. Special versions of some of these runtime base images are referred to as Source-to-Image (S2I) images. With S2I images, you can insert your code into a base image environment that is ready to run that code.
S2I images include:
- .NET
- Java
- Go
- Node.js
- Perl
- PHP
- Python
- Ruby
S2I images are available for you to use directly from the OpenShift Container Platform web console by following procedure:
- Log in to the OpenShift Container Platform web console using your login credentials. The default view for the OpenShift Container Platform web console is the Administrator perspective.
- Use the perspective switcher to switch to the Developer perspective.
- In the +Add view, use the Project drop-down list to select an existing project or create a new project.
- Click All services in the Developer Catalog tile.
- Click Builder Images under Type to see the available S2I images.
S2I images are also available though the Cluster Samples Operator.
12.2.1. Source-to-image build process overview
Source-to-image (S2I) produces ready-to-run images by injecting source code into a container that prepares that source code to be run. It performs the following steps:
-
Runs the
FROM <builder image>
command - Copies the source code to a defined location in the builder image
- Runs the assemble script in the builder image
- Sets the run script in the builder image as the default command
Buildah then creates the container image.
12.2.2. Additional resources
12.3. Customizing source-to-image images
Source-to-image (S2I) builder images include assemble and run scripts, but the default behavior of those scripts is not suitable for all users. You can customize the behavior of an S2I builder that includes default scripts.
12.3.1. Invoking scripts embedded in an image
Builder images provide their own version of the source-to-image (S2I) scripts that cover the most common use-cases. If these scripts do not fulfill your needs, S2I provides a way of overriding them by adding custom ones in the .s2i/bin
directory. However, by doing this, you are completely replacing the standard scripts. In some cases, replacing the scripts is acceptable, but, in other scenarios, you can run a few commands before or after the scripts while retaining the logic of the script provided in the image. To reuse the standard scripts, you can create a wrapper script that runs custom logic and delegates further work to the default scripts in the image.
Procedure
Look at the value of the
io.openshift.s2i.scripts-url
label to determine the location of the scripts inside of the builder image:$ podman inspect --format='{{ index .Config.Labels "io.openshift.s2i.scripts-url" }}' wildfly/wildfly-centos7
Example output
image:///usr/libexec/s2i
You inspected the
wildfly/wildfly-centos7
builder image and found out that the scripts are in the/usr/libexec/s2i
directory.Create a script that includes an invocation of one of the standard scripts wrapped in other commands:
.s2i/bin/assemble
script#!/bin/bash echo "Before assembling" /usr/libexec/s2i/assemble rc=$? if [ $rc -eq 0 ]; then echo "After successful assembling" else echo "After failed assembling" fi exit $rc
This example shows a custom assemble script that prints the message, runs the standard assemble script from the image, and prints another message depending on the exit code of the assemble script.
ImportantWhen wrapping the run script, you must use
exec
for invoking it to ensure signals are handled properly. The use ofexec
also precludes the ability to run additional commands after invoking the default image run script..s2i/bin/run
script#!/bin/bash echo "Before running application" exec /usr/libexec/s2i/run