Chapter 4. Embedding MicroShift applications tutorial
The following tutorial gives a detailed example of how to embed applications in a RHEL for Edge image for use in a MicroShift cluster in various environments.
4.1. Embed application RPMs tutorial
The following tutorial reviews the MicroShift installation steps and adds a description of the workflow for embedding applications. If you are already familiar with rpm-ostree
systems such as Red Hat Enterprise Linux for Edge (RHEL for Edge) and MicroShift, you can go straight to the procedures.
4.1.1. Installation workflow review
Embedding applications requires a similar workflow to embedding MicroShift into a RHEL for Edge image.
- The following image shows how system artifacts such as RPMs, containers, and files are added to a blueprint and used by the image composer to create an ostree commit.
- The ostree commit then can follow either the ISO path or the repository path to edge devices.
- The ISO path can be used for disconnected environments, while the repository path is often used in places were the network is usually connected.
Embedding MicroShift workflow
Reviewing these steps can help you understand the steps needed to embed an application:
- To embed MicroShift on RHEL for Edge, you added the MicroShift repositories to Image Builder.
- You created a blueprint that declared all the RPMs, container images, files and customizations you needed, including the addition of MicroShift.
-
You added the blueprint to Image Builder and ran a build with the Image Builder CLI tool (
composer-cli
). This step createdrpm-ostree
commits, which were used to create the container image. This image contained RHEL for Edge. -
You added the installer blueprint to Image Builder to create an
rpm-ostree
image (ISO) to boot from. This build contained both RHEL for Edge and MicroShift. - You downloaded the ISO with MicroShift embedded, prepared it for use, provisioned it, then installed it onto your edge devices.
4.1.2. Embed application RPMs workflow
After you have set up a build host that meets the Image Builder requirements, you can add your application in the form of a directory of manifests to the image. After those steps, the simplest way to embed your application or workload into a new ISO is to create your own RPMs that include the manifests. Your application RPMs contain all of the configuration files describing your deployment.
The following "Embedding applications workflow" image shows how Kubernetes application manifests and RPM spec files are combined in a single application RPM build. This build becomes the RPM artifact included in the workflow for embedding MicroShift in an ostree commit.
Embedding applications workflow
The following procedures use the rpmbuild
tool to create a specification file and local repository. The specification file defines how the package is built, moving your application manifests to the correct location inside the RPM package for MicroShift to pick them up. That RPM package is then embedded in the ISO.
4.1.3. Preparing to make application RPMs
To build your own RPMs, choose a tool of your choice, such as the rpmbuild
tool, and initialize the RPM build tree in your home directory. The following is an example procedure. As long as your RPMs are accessible to Image Builder, you can use the method you prefer to build the application RPMs.
Prerequisites
- You have set up a Red Hat Enterprise Linux for Edge (RHEL for Edge) 9.4 build host that meets the Image Builder system requirements.
- You have root access to the host.
Procedure
Install the
rpmbuild
tool and create the yum repository for it by running the following command:$ sudo dnf install rpmdevtools rpmlint yum-utils createrepo
Create the file tree you need to build RPM packages by running the following command:
$ rpmdev-setuptree
Verification
List the directories to confirm creation by running the following command:
$ ls ~/rpmbuild/
Example output
BUILD RPMS SOURCES SPECS SRPMS
4.1.4. Building the RPM package for the application manifests
To build your own RPMs, you must create a spec file that adds the application manifests to the RPM package. The following is an example procedure. As long as the application RPMs and other elements needed for image building are accessible to Image Builder, you can use the method that you prefer.
Prerequisites
- You have set up a Red Hat Enterprise Linux for Edge (RHEL for Edge) 9.4 build host that meets the Image Builder system requirements.
- You have root access to the host.
- The file tree required to build RPM packages was created.
Procedure
In the
~/rpmbuild/SPECS
directory, create a file such as<application_workload_manifests.spec>
using the following template:Example spec file
Name: <application_workload_manifests> Version: 0.0.1 Release: 1%{?dist} Summary: Adds workload manifests to microshift BuildArch: noarch License: GPL Source0: %{name}-%{version}.tar.gz #Requires: microshift %description Adds workload manifests to microshift %prep %autosetup %install 1 rm -rf $RPM_BUILD_ROOT mkdir -p $RPM_BUILD_ROOT/%{_prefix}/lib/microshift/manifests cp -pr ~/manifests $RPM_BUILD_ROOT/%{_prefix}/lib/microshift/ %clean rm -rf $RPM_BUILD_ROOT %files %{_prefix}/lib/microshift/manifests/** %changelog * <DDD MM DD YYYY username@domain - V major.minor.patch> - <your_change_log_comment>
- 1
- The
%install
section creates the target directory inside the RPM package,/usr/lib/microshift/manifests/
and copies the manifests from the source home directory,~/manifests
.
ImportantAll of the required YAML files must be in the source home directory
~/manifests
, including akustomize.yaml
file if you are using kustomize.Build your RPM package in the
~/rpmbuild/RPMS
directory by running the following command:$ rpmbuild -bb ~/rpmbuild/SPECS/<application_workload_manifests.spec>
4.1.5. Adding application RPMs to a blueprint
To add application RPMs to a blueprint, you must create a local repository that Image Builder can use to create the ISO. With this procedure, the required container images for your workload can be pulled over the network.
Prerequisites
- You have root access to the host.
-
Workload or application RPMs exist in the
~/rpmbuild/RPMS
directory.
Procedure
Create a local RPM repository by running the following command:
$ createrepo ~/rpmbuild/RPMS/
Give Image Builder access to the RPM repository by running the following command:
$ sudo chmod a+rx ~
NoteYou must ensure that Image Builder has all of the necessary permissions to access all of the files needed for image building, or the build cannot proceed.
Create the blueprint file,
repo-local-rpmbuild.toml
using the following template:id = "local-rpm-build" name = "RPMs build locally" type = "yum-baseurl" url = "file://<path>/rpmbuild/RPMS" 1 check_gpg = false check_ssl = false system = false
- 1
- Specify part of the path to create a location that you choose. Use this path in the later commands to set up the repository and copy the RPMs.
Add the repository as a source for Image Builder by running the following command:
$ sudo composer-cli sources add repo-local-rpmbuild.toml
Add the RPM to your blueprint, by adding the following lines:
… [[packages]] name = "<application_workload_manifests>" 1 version = "*" …
- 1
- Add the name of your workload here.
Push the updated blueprint to Image Builder by running the following command:
$ sudo composer-cli blueprints push repo-local-rpmbuild.toml
At this point, you can either run Image Builder to create the ISO, or embed the container images for offline use.
To create the ISO, start Image Builder by running the following command:
$ sudo composer-cli compose start-ostree repo-local-rpmbuild edge-commit
In this scenario, the container images are pulled over the network by the edge device during startup.
4.2. Additional resources
- Embedding applications for offline use
- Embedding MicroShift in an RPM-OSTree image
- Composing, installing, and managing RHEL for Edge images
- Preparing for image building
- Meet Red Hat Device Edge with MicroShift
- How to create a Linux RPM package
- Composing a RHEL for Edge image using image builder command-line
- Image Builder system requirements