Search

Chapter 10. Managing RHEL for Edge images

download PDF

To manage the RHEL for Edge images, you can perform any of the following administrative tasks:

  • Edit the RHEL for Edge image blueprint by using image builder in RHEL web console or in the command-line
  • Build a commit update by using image builder command-line
  • Update the RHEL for Edge images
  • Configure rpm-ostree remotes on nodes, to update node policy
  • Restore RHEL for Edge images manually or automatically by using greenboot

10.1. Editing a RHEL for Edge image blueprint by using image builder

You can edit the RHEL for Edge image blueprint to:

  • Add additional components that you might require
  • Modify the version of any existing component
  • Remove any existing component

10.1.1. Adding a component to RHEL for Edge blueprint using image builder in RHEL web console

To add a component to a RHEL for Edge image blueprint, ensure that you have met the following prerequisites and then follow the procedure to edit the corresponding blueprint.

Prerequisites

  • On a RHEL system, you have accessed the RHEL image builder dashboard.
  • You have created a blueprint for RHEL for Edge image.

Procedure

  1. On the RHEL image builder dashboard, click the blueprint that you want to edit.

    To search for a specific blueprint, enter the blueprint name in the filter text box, and click Enter.

  2. On the upper right side of the blueprint, click Edit Packages.

    The Edit blueprints wizard opens.

  3. On the Details page, update the blueprint name and click Next.
  4. On the Packages page, follow the steps:

    1. In the Available Packages, enter the package name that you want to add in the filter text box, and click Enter.

      A list with the component name appears.

    2. Click > to add the component to the blueprint.
  5. On the Review page, click Save.

    The blueprint is now updated with the new package.

10.1.2. Removing a component from a blueprint using RHEL image builder in the web console

To remove one or more unwanted components from a blueprint that you created by using RHEL image builder, ensure that you have met the following prerequisites and then follow the procedure.

Prerequisites

  • On a RHEL system, you have accessed the RHEL image builder dashboard.
  • You have created a blueprint for RHEL for Edge image.
  • You have added at least one component to the RHEL for Edge blueprint.

Procedure

  1. On the RHEL image builder dashboard, click the blueprint that you want to edit.

    To search for a specific blueprint, enter the blueprint name in the filter text box, and click Enter.

  2. On the upper right side of the blueprint, click Edit Packages.

    The Edit blueprints wizard opens.

  3. On the Details page, update the blueprint name and click Next.
  4. On the Packages page, follow the steps:

    1. From the Chosen packages, click < to remove the chosen component. You can also click << to remove all the packages at once.
  5. On the Review page, click Save.

    The blueprint is now updated.

10.1.3. Editing a RHEL for Edge image blueprint using command-line interface

You can change the specifications for your RHEL for Edge image blueprint by using RHEL image builder command-line. To do so, ensure that you have met the following prerequisites and then follow the procedure to edit the corresponding blueprint.

Prerequisites

  • You have access to the RHEL image builder command-line.
  • You have created a RHEL for Edge image blueprint.

Procedure

  1. Save (export) the blueprint to a local text file:

    # composer-cli blueprints save BLUEPRINT-NAME
  2. Edit the BLUEPRINT-NAME.toml file with a text editor of your choice and make your changes.

    Before finishing with the edits, verify that the file is a valid blueprint:

  3. Increase the version number.

    Ensure that you use a Semantic Versioning scheme.

    Note

    if you do not change the version, the patch component of the version is increased automatically.

  4. Check if the contents are valid TOML specifications. See the TOML documentation for more information.

    Note

    TOML documentation is a community product and is not supported by Red Hat. You can report any issues with the tool at https://github.com/toml-lang/toml/issues.

  5. Save the file and close the editor.
  6. Push (import) the blueprint back into RHEL image builder server:

    # composer-cli blueprints push BLUEPRINT-NAME.toml
    Note

    When pushing the blueprint back into the RHEL image builder server, provide the file name including the .toml extension.

  7. Verify that the contents uploaded to RHEL image builder match your edits:

    # composer-cli blueprints show BLUEPRINT-NAME
  8. Check whether the components and versions listed in the blueprint and their dependencies are valid:

    # composer-cli blueprints depsolve BLUEPRINT-NAME

10.2. Updating RHEL for Edge images

10.2.1. How RHEL for Edge image updates are deployed

With RHEL for Edge images, you can either deploy the updates manually or can automate the deployment process. The updates are applied in an atomic manner, where the state of each update is known, and the updates are staged and applied only upon reboot. Because no changes are seen until you reboot the device, you can schedule a reboot to ensure the highest possible uptime.

During the image update, only the updated operating system content is transferred over the network. This makes the deployment process more efficient compared to transferring the entire image. The operating system binaries and libraries in /usr are read-only, and the read and write state is maintained in /var and /etc directories.

When moving to a new deployment, the /etc and the /var directories are copied to the new deployment with read and write permissions. The /usr directory is copied as a soft link to the new deployment directory, with read-only permissions.

The following diagram illustrates the RHEL for Edge image update deployment process:

Image Deployment

By default, the new system is booted using a procedure similar to a chroot operation, that is, the system enables control access to a filesystem while controlling the exposure to the underlying server environment. The new /sysroot directory mainly has the following parts:

  • Repository database at the /sysroot/ostree/repo directory.
  • File system revisions at the /sysroot/ostree/deploy/rhel/deploy directory, which are created by each operation in the system update.
  • The /sysroot/ostree/boot directory, which links to deployments on the previous point. Note that /ostree is a soft link to /sysroot/ostree. The files from the /sysroot/ostree/boot directory are not duplicated. The same file is used if it is not changed during the deployment. The files are hard-links to another file stored in the /sysroot/ostree/repo/objects directory.

The operating system selects the deployment in the following way:

  1. The dracut tool parses the ostree kernel argument in the initramfs root file system and sets up the /usr directory as a read-only bind mount.
  2. Bind the deployment directory in /sysroot to / directory.
  3. Re-mount the operating system already mounted dirs using the MS_MOVE mount flag

If anything goes wrong, you can perform a deployment rollback by removing the old deployments with the rpm-ostree cleanup command. Each client machine contains an OSTree repository stored in /ostree/repo, and a set of deployments stored in /ostree/deploy/$STATEROOT/$CHECKSUM.

With the deployment updates in RHEL for Edge image, you can benefit from a better system consistency across multiple devices, easier reproducibility, and better isolation between the pre and post system states change.

10.2.2. Building a commit update

You can build a commit update after making a change in the blueprint, such as:

  • Adding an additional package that your system requires
  • Modifying the package version of any existing component
  • Removing any existing package.

Prerequisites

Procedure

  1. Start the compose of the new commit image, with the following arguments: --url, --ref, blueprint-name, edge-commit.

    # composer-cli compose start-ostree --ref rhel/8/x86_64/edge --url http://localhost:8080/repo <blueprint-name> edge-commit

    The command instructs the compose process to fetch the metadata from the OStree repo before starting the compose. The resulting new OSTree commit contains a reference of the original OSTree commit as a parent image.

  2. After the compose process finishes, fetch the .tar file.

    # composer-cli compose image <UUID>
  3. Extract the commit to a temporary directory, so that you can store the commit history in the OSTree repo.

    $ tar -xf UUID.tar -C /var/tmp
  4. Inspect the resulting OSTree repo commit, by using the tar -xf command. It extracts the tar file to disk so you can inspect the resulting OSTree repo:

    $ ostree --repo=/var/tmp/repo log rhel/8/x86_64/edge
    commit d523ef801e8b1df69ddbf73ce810521b5c44e9127a379a4e3bba5889829546fa
    Parent:  f47842de7e6859cee07d743d3c67949420874727883fa9dbbaeb5824ad949d91
    ContentChecksum:  f0f6703696331b661fa22d97358db48ba5f8b62711d9db83a00a79b3ae0dfe16
    Date:  2023-06-04 20:22:28 /+0000
    Version: 8

    In the output example, there is a single OSTree commit in the repo that references a parent commit. The parent commit is the same checksum from the original OSTree commit that you previously made.

  5. Merge the two commits by using the ostree pull-local command:

    $ sudo ostree --repo=/var/srv/httpd/repo pull-local /var/tmp/repo
    20 metadata, 22 content objects imported; 0 bytes content written

    This command copies any new metadata and content from the location on the disk, for example, /var/tmp, to a destination OSTree repo in /var/srv/httpd.

Verification

  1. Inspect the target OSTree repo:

    $ ostree --repo=/var/srv/httpd/repo log rhel/8/x86_64/edge
    commit d523ef801e8b1df69ddbf73ce810521b5c44e9127a379a4e3bba5889829546fa
    Parent:  f47842de7e6859cee07d743d3c67949420874727883fa9dbbaeb5824ad949d91
    ContentChecksum:  f0f6703696331b661fa22d97358db48ba5f8b62711d9db83a00a79b3ae0dfe16
    Date:  2023-06-04 20:22:28 /+0000
    Version: 8
    (no subject)
    
    commit f47842de7e6859cee07d743d3c67949420874727883fa9dbbaeb5824ad949d91
    ContentChecksum:  9054de3fe5f1210e3e52b38955bea0510915f89971e3b1ba121e15559d5f3a63
    Date:  2023-06-04 20:01:08 /+0000
    Version: 8
    (no subject)

    You can see that the target OSTree repo now contains two commits in the repository, in a logical order. After successful verification, you can update your RHEL for Edge system.

10.2.3. Deploying RHEL for Edge image updates manually

After you have edited a RHEL for Edge blueprint, you can update the image commit. RHEL image builder generates a new commit for the updated RHEL for Edge image. Use this new commit to deploy the image with latest package versions or with additional packages.

To deploy RHEL for Edge images updates, ensure that you meet the prerequisites and then follow the procedure.

Prerequisites

  • On a RHEL system, you have accessed the RHEL image builder dashboard.
  • You have created a RHEL for Edge image blueprint.
  • You have edited the RHEL for Edge image blueprint.

Procedure

  1. On the RHEL image builder dashboard click Create Image.
  2. On the Create Image window, perform the following steps:

    1. In the Image output page:

      1. From the Select a blueprint dropdown list, select the blueprint that you edited.
      2. From the Image output type dropdown list, select RHEL for Edge Commit (.tar). Click Next.
    2. In the OSTree settings page, enter:

      1. In the Repository URL field, enter the URL to the OSTree repository of the commit to embed in the image. For example, http://10.0.2.2:8080/repo/. See Setting up a web server to install RHEL for Edge image.
      2. In the Parent commit field, specify the parent commit ID that was previously generated. See Extracting RHEL for Edge image commit.
      3. In the Ref field, you can either specify a name for your commit or leave it empty. By default, the web console specifies the Ref as rhel/8/arch_name/edge. Click Next.
    3. In the Review page, check the customizations and click Create image. RHEL image builder starts to create a RHEL for Edge image for the updated blueprint. The image creation process takes a few minutes to complete.

      To view the RHEL for Edge image creation progress, click the blueprint name from the breadcrumbs, and then click the Images tab.

      The resulting image includes the latest packages that you have added, if any, and have the original commit ID as a parent.

  3. Download the resulting RHEL for Edge Commit (.tar) image.

    1. From the Images tab, click Download to save the RHEL for Edge Commit (.tar) image to your system.
  4. Extract the OSTree commit (.tar) file.

    # tar -xf UUID-commit.tar -C UPGRADE_FOLDER
  5. Upgrade the OSTree repo:

    # ostree --repo=/usr/share/nginx/html/repo pull-local UPGRADE_FOLDER
    # ostree --repo=/usr/share/nginx/html/repo summary -u
  6. On the RHEL system provisioned, from the original edge image, verify the current status.

    $ rpm-ostree status

    If there is no new commit ID, run the following command to verify if there is any upgrade available:

    $ rpm-ostree upgrade --check

    The command output provides the current active OSTree commit ID.

  7. Update OSTree to make the new OSTree commit ID available.

    $ rpm-ostree upgrade

    OSTree verifies if there is an update on the repository. If yes, it fetches the update and requests you to reboot your system so that you can activate the deployment of this new commit update.

  8. Check the current status again:

    $ rpm-ostree status

    You can now see that there are 2 commits available:

    • The active parent commit.
    • A new commit that is not active and contains 1 added difference.
  9. To activate the new deployment and to make the new commit active, reboot your system.

    # systemctl reboot

    The Anaconda Installer reboots into the new deployment. On the login screen, you can see a new deployment available for you to boot.

  10. If you want to boot into the newest deployment (commit), the rpm-ostree upgrade command automatically orders the boot entries so that the new deployment is first in the list. Optionally, you can use the arrow key on your keyboard to select the GRUB menu entry and press Enter.
  11. Provide your login user account credentials.
  12. Verify the OSTree status:

    $ rpm-ostree status

    The command output provides the active commit ID.

  13. To view the changed packages, if any, run a diff between the parent commit and the new commit:

    $ rpm-ostree db diff parent_commit new_commit

    The update shows that the package you have installed is available and ready for use.

10.2.4. Deploying RHEL for Edge image updates manually using the command-line

After you have edited a RHEL for Edge blueprint, you can update the image commit. RHEL image builder generates a new commit for the updated RHEL for Edge image. Use the new commit to deploy the image with latest package versions or with additional packages using the CLI.

To deploy RHEL for Edge image updates using the CLI, ensure that you meet the prerequisites, and then follow the procedure.

Prerequisites

Procedure

  1. Create the RHEL for Edge Commit (.tar) image with the following arguments:

    # composer-cli compose start-ostree --ref ostree_ref --url URL-OSTree-repository -blueprint_name_ image-type

    where

    • ref is the reference you provided during the creation of the RHEL for Edge Container commit. For example, rhel/8/x86_64/edge.
    • URL-OSTree-repository is the URL to the OSTree repository of the commit to embed in the image. For example, http://10.0.2.2:8080/repo/. See Setting up a web server to install RHEL for Edge image.
    • image-type is edge-commit.

      RHEL image builder creates a RHEL for Edge image for the updated blueprint.

  2. Check the RHEL for Edge image creation progress:

    # composer-cli compose status
    Note

    The image creation processes can take up to ten to thirty minutes to complete.

    The resulting image includes the latest packages that you have added, if any, and has the original commit ID as a parent.

  3. Download the resulting RHEL for Edge image. For more information, see Downloading a RHEL for Edge image using the RHEL image builder command-line interface.
  4. Extract the OSTree commit.

    # tar -xf UUID-commit.tar -C upgrade_folder
  5. Serve the OSTree commit by using httpd. See Setting up a web server to install RHEL for Edge image.
  6. Upgrade the OSTree repo:

    # ostree --repo=/var/www/html/repo pull-local /tmp/ostree-commit/repo
    # ostree --repo=/var/www/html/repo summary -u
  7. On the RHEL system provisioned from the original edge image, verify the current status:

    $ rpm-ostree status

    If there is no new commit ID, run the following command to verify if there is any upgrade available:

    $ rpm-ostree upgrade --check

    The command output provides the current active OSTree commit ID.

  8. Update OSTree to make the new OSTree commit ID available:

    $ rpm-ostree upgrade

    OSTree verifies if there is an update on the repository. If yes, it fetches the update and requests you to reboot your system so that you can activate the deployment of the new commit update.

  9. Check the current status again:

    $ rpm-ostree status

    You should now see that there are 2 commits available:

    • The active parent commit
    • A new commit that is not active and contains 1 added difference
  10. To activate the new deployment and make the new commit active, reboot your system:

    # systemctl reboot

    The Anaconda Installer reboots into the new deployment. On the login screen, you can see a new deployment available for you to boot.

  11. If you want to boot into the newest deployment, the rpm-ostree upgrade command automatically orders the boot entries so that the new deployment is first in the list. Optionally, you can use the arrow key on your keyboard to select the GRUB menu entry and press Enter.
  12. Log in using your account credentials.
  13. Verify the OSTree status:

    $ rpm-ostree status

    The command output provides the active commit ID.

  14. To view the changed packages, if any, run a diff between the parent commit and the new commit:

    $ rpm-ostree db diff parent_commit new_commit

    The update shows that the package you have installed is available and ready for use.

10.2.5. Deploying RHEL for Edge image updates manually for non-network-base deployments

After editing a RHEL for Edge blueprint, you can update your RHEL for Edge Commit image with those updates. Use RHEL image builder to generate a new commit to update your RHEL for Edge image that is already deployed in a VM, for example. Use this new commit to deploy the image with latest package versions or with additional packages.

To deploy RHEL for Edge images updates, ensure that you meet the prerequisites and then follow the procedure.

Prerequisites

  • On your host, you have opened the RHEL image builder app from the web console in a browser.
  • You have a RHEL for Edge system provisioned that is up and running.
  • You have an OSTree repository that is being served over HTTP.
  • You have edited a previously created RHEL for Edge image blueprint.

Procedure

  1. On your system host, on the RHEL image builder dashboard, click Create Image.
  2. On the Create Image window, perform the following steps:

    1. In the Image output page:

      1. From the Select a blueprint dropdown list, select the blueprint that you edited.
      2. From the Image output type dropdown list, select RHEL for Edge Container (.tar).
      3. Click Next.
    2. In the OSTree settings page, enter:

      1. In the Repository URL field, enter the URL to the OSTree repository of the commit to embed in the image. For example, http://10.0.2.2:8080/repo/. See Setting up a web server to install RHEL for Edge image.
      2. In the Parent commit field, specify the parent commit ID that was previously generated. See Extracting RHEL for Edge image commit.
      3. In the Ref field, you can either specify a name for your commit or leave it empty. By default, the web console specifies the Ref as rhel/8/arch_name/edge.
      4. Click Next.
    3. In the Review page, check the customizations and click Create.

      RHEL image builder creates a RHEL for Edge image for the updated blueprint.

    4. Click the Images tab to view the progress of RHEL for Edge image creation.

      Note

      The image creation process takes a few minutes to complete.

      The resulting image includes the latest packages that you have added, if any, and has the original commit ID as a parent.

  3. Download the resulting RHEL for Edge image on your host:

    1. From the Images tab, click Download to save the RHEL for Edge Container (.tar) image to your host system.
  4. On the RHEL system provisioned from the original edge image, perform the following steps:

    1. Load the RHEL for Edge Container image into Podman, serving the child commit ID this time.

      $ cat ./child-commit_ID-container.tar | sudo podman load
    2. Run Podman.

      #  sudo podman run -p 8080:8080 localhost/edge-test
    3. Upgrade the OSTree repo:

      # ostree --repo=/var/www/html/repo pull-local /tmp/ostree-commit/repo
      # ostree --repo=/var/www/html/repo summary -u
    4. On the RHEL system provisioned, from the original edge image, verify the current status.

      $ rpm-ostree status

      If there is no new commit ID, run the following command to verify if there is any upgrade available:

      $ rpm-ostree upgrade --check

      If there are updates available, the command output provides information about the available updates in the OSTree repository, such as the current active OSTree commit ID. Else, it prompts a message informing that there are no updates available.

    5. Update OSTree to make the new OSTree commit ID available.

      $ rpm-ostree upgrade

      OSTree verifies if there is an update on the repository. If yes, it fetches the update and requests you to reboot your system so that you can activate the deployment of this new commit update.

    6. Check the current system status:

      $ rpm-ostree status

      You can now see that there are 2 commits available:

      • The active parent commit.
      • A new commit that is not active and contains 1 added difference.
    7. To activate the new deployment and to make the new commit active, reboot your system.

      # systemctl reboot

      The Anaconda Installer reboots into the new deployment. On the login screen, you can see a new deployment available for you to boot.

    8. To boot into the newest commit, run the following command to automatically order the boot entries so that the new deployment is first in the list:

      $ rpm-ostree upgrade

      Optionally, you can use the arrow key on your keyboard to select the GRUB menu entry and press Enter.

  5. Provide your login user account credentials.
  6. Verify the OSTree status:

    $ rpm-ostree status

    The command output provides the active commit ID.

  7. To view the changed packages, if any, run a diff between the parent commit and the new commit:

    $ rpm-ostree db diff parent_commit new_commit

    The update shows that the package you have installed is available and ready for use.

10.3. Deploying RHEL for Edge automatic image updates

After you install a RHEL for Edge image on an Edge device, you can check for image updates available, if any, and can auto-apply them.

The rpm-ostreed-automatic.service (systemd service) and rpm-ostreed-automatic.timer (systemd timer) control the frequency of checks and upgrades. The available updates, if any, appear as staged deployments.

Deploying automatic image updates involves the following high-level steps:

  • Update the image update policy
  • Enable automatic download and staging of updates

10.3.1. Updating the RHEL for Edge image update policy

To update the image update policy, use the AutomaticUpdatePolicy and an IdleExitTimeout setting from the rpm-ostreed.conf file at /etc/rpm-ostreed.conf location on an Edge device.

The AutomaticUpdatePolicy settings controls the automatic update policy and has the following update checks options:

  • none: Disables automatic updates. By default, the AutomaticUpdatePolicy setting is set to none.
  • check: Downloads enough metadata to display available updates with rpm-ostree status.
  • stage: Downloads and unpacks the updates that are applied on a reboot.

The IdleExitTimeout setting controls the time in seconds of inactivity before the daemon exit and has the following options:

  • 0: Disables auto-exit.
  • 60: By default, the IdleExitTimeout setting is set to 60.

To enable automatic updates, perform the following steps:

Procedure

  1. In the /etc/rpm-ostreed.conf file, update the following:

    • Change the value of AutomaticUpdatePolicy to check.
    • To run the update checks, specify a value in seconds for IdleExitTimeout.
  2. Reload the rpm-ostreed service and enable the systemd timer.

    # systemctl reload rpm-ostreed
    # systemctl enable rpm-ostreed-automatic.timer --now
  3. Verify the rpm-ostree status to ensure the automatic update policy is configured and time is active.

    # rpm-ostree status

    The command output shows the following:

    State: idle; auto updates enabled (check; last run <minutes> ago)

    Additionally, the output also displays information about the available updates.

10.3.2. Enabling RHEL for Edge automatic download and staging of updates

After you update the image update policy to check for image updates, the updates if any are displayed along with the update details. If you decide to apply the updates, enable the policy to automatically download and stage the updates. The available image updates are then downloaded and staged for deployment. The updates are applied and take effect when you reboot the Edge device.

To enable the policy for automatic download and staging of updates, perform the following updates:

Procedure

  1. In the /etc/rpm-ostreed.conf file, update "AutomaticUpdatePolicy" to stage.
  2. Reload the rpm-ostreed service.

    # systemctl enable rpm-ostreed-automatic.timer --now
  3. Verify the rpm-ostree status

    # rpm-ostree status

    The command output shows the following:

    State: idle
    AutomaticUpdates: stage; rpm-ostreed-automatic.timer: last run <time> ago
  4. To initiate the updates, you can either wait for the timer to initiate the updates, or can manually start the service.

    # systemctl start rpm-ostreed-automatic.service

    After the updates are initiated, the rpm-ostree status shows the following:

    # rpm-ostree status
    State: busy
    AutomaticUpdates: stage; rpm-ostreed-automatic.service: running
    Transaction: automatic (stage)

    When the update is complete, a new deployment is staged in the list of deployments, and the original booted deployment is left untouched. You can decide if you want to boot the system using the new deployment or can wait for the next update.

    To view the list of deployments, run the rpm-ostree status command.

    Following is a sample output.

    # rpm-ostree status
    State: idle
    AutomaticUpdates: stage; rpm-ostreed-automatic.timer: last run <time> ago
    Deployments:

    To view the list of deployments with the updated package details, run the rpm-ostree status -v command.

10.4. Rolling back RHEL for Edge images

Because RHEL for Edge applies transactional updates to the operating system, you can either manually or automatically roll back the unsuccessful updates to the last known good state, which prevents system failure during updates. You can automate the verification and rollback process by using the greenboot framework.

The greenboot health check framework leverages rpm-ostree to run custom health checks on system startup. In case of an issue, the system rolls back to the last working state. When you deploy a rpm-ostree update, it runs scripts to check that critical services can still work after the update. If the system does not work, for example, due to some failed package, you can roll back the system to a previous stable version of the system. This process ensures that your RHEL for Edge device is in an operational state.

After you update an image, it creates a new image deployment while preserving the previous image deployment. You can verify whether the update was successful. If the update is unsuccessful, for example, due to a failed package, you can roll back the system to a previous stable version.

10.4.1. Introducing the greenboot checks

Greenboot is a Generic Health Check Framework for systemd available on rpm-ostree based systems. It contains the following RPM packages that you can install on your system:

  • greenboot - a package that contains the following functionalities:

    • Checking provided scripts
    • Reboot the system if the check fails
    • Rollback to a previous deployment the reboot did not solve the issue.
  • greenboot-default-health-checks - a set of optional and selected health checks provided by your greenboot system maintainers.

Greenboot works in a RHEL for Edge system by using health check scripts that run on the system to assess the system health and automate a rollback to the last healthy state in case of some software fails. These health checks scripts are available in the /etc/greenboot/check/required.d directory. Greenboot supports shell scripts for the health checks. Having a health check framework is especially useful when you need to check for software problems and perform system rollbacks on edge devices where direct serviceability is either limited or non-existent. When you install and configure health check scripts, it triggers the health checks to run every time the system starts.

You can create your own health check scripts to assess the health of your workloads and applications. These additional health check scripts are useful components of software problem checks and automatic system rollbacks.

Note

You cannot use rollback in case of any health check failure on a system that is not using OSTree.

10.4.2. RHEL for Edge images roll back with greenboot

With RHEL for Edge images, only transactional updates are applied to the operating system. The transactional updates are atomic, which means that the updates are applied only if all the updates are successful, and there is support for rollbacks. With the transactional updates, you can easily rollback the unsuccessful updates to the last known good state, preventing system failure during updates.

Performing health checks is especially useful when you need to check for software problems and perform system rollbacks on edge devices where direct serviceability is limited or non-existent.

Note

You cannot use rollback in case of an update failure on a system that is not using OSTree, even if health checks might run.

You can use intelligent rollbacks with the greenboot health check framework to automatically assess system health every time the system starts. You can obtain pre-configured health from the greenboot-default-health-checks subpackage. These checks are located in the /usr/lib/greenboot/check read-only directory in rpm-ostree systems.

Greenboot leverages rpm-ostree and runs custom health checks that run on system startup. In case of an issue, the system rolls back the changes and preserves the last working state. When you deploy an rpm-ostree update, it runs scripts to check that critical services can still work after the update. If the system does not work, the update rolls back to the last known working version of the system. This process ensures that your RHEL for Edge device is in an operational state.

You can obtain pre-configured health from the greenboot-default-health-checks`subpackage. These checks are located in the `/usr/lib/greenboot/check read-only directory in rpm-ostree systems. You can also configure shell scripts as the following types of checks:

Example 10.1. The greenboot directory structure

etc
└─ greenboot
   ├─ check
   |   └─ required.d
   |   └─ init.py
   └─ green.d
   └─ red.d
Required
Contains the health checks that must not fail. Place required shell scripts in the /etc/greenboot/check/required.d directory. If the scripts fail, greenboot retries them three times by default. You can configure the number of retries in the /etc/greenboot/greenboot.conf file by setting the GREENBOOT_MAX_BOOTS parameter to the number of retries you want.

After all retries fail, greenboot automatically initiates a rollback if one is available. If a rollback is not available, the system log output shows that you need to perform a manual intervention.

Wanted
Contains the health checks that might fail without causing the system to roll back. Place wanted shell scripts in the /etc/greenboot/check/wanted.d directory. Greenboot informs that the script fails, the system health status remains unaffected and it does not perform a rollback neither a reboot.

You can also specify shell scripts that will run after a check:

Green
Contains the scripts to run after a successful boot. Place these scripts into the /etc/greenboot/green.d`directory. Greenboot informs that the boot was successful.
Red
Contains the scripts to run after a failed boot. Place these scripts into the /etc/greenboot/red.d directory. The system attempts to boot three times and in case of failure, it executes the scripts. Greenboot informs that the boot failed.

The following diagram illustrates the RHEL for Edge image roll back process.

Image restoration process

After booting the updated operating system, greenboot runs the scripts in the required.d and wanted.d directories. If any of the scripts fail in the required.d directory, greenboot runs any scripts in the red.d directory, and then reboots the system.

Greenboot makes 2 more attempts to boot on the upgraded system. If during the third boot attempt the scripts in required.d are still failing, greenboot runs the red.d scripts one last time, to ensure that the script in the red.d directory tried to make a corrective action to fix the issue and this was not successful. Then, greenboot rollbacks the system from the current rpm-ostree deployment to the previous stable deployment.

10.4.3. Greenboot health check status

When deploying your updated system, wait until the greenboot health checks have finished before making the changes to ensure that those changes are not lost if greenboot rolls the system back to an earlier state. If you want to make configuration changes or deploy applications you must wait until the greenboot health checks have finished. This ensures that your changes are not lost if greenboot rolls your rpm-ostree system back to an earlier state.

The greenboot-healthcheck service runs once and then exits. You can check the status of the service to know if it is done, and to know the outcome, by using the following commands:

systemctl is-active greenboot-healthcheck.service
This command reports active when the service has exited. If it the service did not even run it shows inactive.
systemctl show --property=SubState --value greenboot-healthcheck.service
Reports exited when done, running while still running.
systemctl show --property=Result --value greenboot-healthcheck.service
Reports success when the checks passed.
systemctl show --property=ExecMainStatus --value greenboot-healthcheck.service
Reports the numerical exit code of the service, 0 means success and nonzero values mean a failure occurred.
cat /run/motd.d/boot-status
Shows a message, such as "Boot Status is GREEN - Health Check SUCCESS".

10.4.4. Checking greenboot health checks statuses

Check the status of greenboot health checks before making changes to the system or during troubleshooting. You can use any of the following commands to help you ensure that greenboot scripts have finished running.

  • Use one of the following options to check the statuses:

    • To see a report of health check status, enter:

      $ systemctl show --property=SubState --value greenboot-healthcheck.service

      The following outputs are possible:

      • start means that greenboot checks are still running.
      • exited means that checks have passed and greenboot has exited. Greenboot runs the scripts in the green.d directory when the system is in a healthy state.
      • failed means that checks have not passed. Greenboot runs the scripts in red.d directory when the system is in this state and might restart the system.
    • To see a report showing the numerical exit code of the service, where 0 means success and nonzero values mean a failure occurred, use the following command:

      $ systemctl show --property=ExecMainStatus --value greenboot-healthcheck.service
    • To see a report showing a message about boot status, such as Boot Status is GREEN - Health Check SUCCESS, enter:

      $ cat /run/motd.d/boot-status

10.4.5. Manually rolling back RHEL for Edge images

When you upgrade your operating system, a new deployment is created, and the rpm-ostree package also keeps the previous deployment. If there are issues on the updated version of the operating system, you can manually roll back to the previous deployment with a single rpm-ostree command, or by selecting the previous deployment in the GRUB boot loader.

To manually roll back to a previous version, perform the following steps.

Prerequisite

  1. You updated your system and it failed.

Procedure

  1. Optional: Check for the fail error message:

    $ journalctl -u greenboot-healthcheck.service.
  2. Run the rollback command:

    # rpm-ostree rollback

    The command output provides details about the commit ID that is being moved and indicates a completed transaction with the details of the package being removed.

  3. Reboot the system.

    # systemctl reboot

    The command activates the previous commit with the stable content. The changes are applied and the previous version is restored.

10.4.6. Rolling back RHEL for Edge images using an automated process

Greenboot checks provides a framework that is integrated into the boot process and can trigger rpm-ostree rollbacks when a health check fails. For the health checks, you can create a custom script that indicates whether a health check passed or failed. Based on the result, you can decide when a rollback should be triggered. The following procedure shows how to create an health check script example:

Procedure

  1. Create a script that returns a standard exit code 0.

    For example, the following script ensures that the configured DNS server is available:

    #!/bin/bash
    
    DNS_SERVER=$(grep ^nameserver /etc/resolv.conf | head -n 1 | cut -f2 -d" ")
    COUNT=0
    # check DNS server is available
    ping -c1 $DNS_SERVER
    while [ $? != '0' ] && [ $COUNT -lt 10 ]; do
    ((COUNT++))
    echo "Checking for DNS: Attempt $COUNT ."
    sleep 10
    ping -c 1 $DNS_SERVER
    done
  2. Include an executable file for the health checks at /etc/greenboot/check/required.d/.

    chmod +x check-dns.sh

    During the next reboot, the script is executed as part of the boot process, before the system enters the boot-complete.target unit. If the health checks are successful, no action is taken. If the health checks fail, the system will reboot several times, before marking the update as failed and rolling back to the previous update.

Verification steps

To check if the default gateway is reachable, run the following health check script:

  1. Create a script that returns a standard exit code 0.

    #!/bin/bash
    
    DEF_GW=$(ip r | awk '/^default/ {print $3}')
    SCRIPT=$(basename $0)
    
    count=10
    connected=0
    ping_timeout=5
    interval=5
    
    while [ $count -gt 0 -a $connected -eq 0 ]; do
      echo "$SCRIPT: Pinging default gateway $DEF_GW"
      ping -c 1 -q -W $ping_timeout $DEF_GW > /dev/null 2>&1 && connected=1 || sleep $interval
      ((--count))
    done
    
    if [ $connected -eq 1 ]; then
      echo "$SCRIPT: Default gateway $DEF_GW is reachable."
      exit 0
    else
      echo "$SCRIPT: Failed to ping default gateway $DEF_GW!" 1>&2
      exit 1
    fi
  2. Include an executable file for the health checks at /etc/greenboot/check/required.d/ directory.

    chmod +x check-gw.sh
Red Hat logoGithubRedditYoutubeTwitter

Learn

Try, buy, & sell

Communities

About Red Hat Documentation

We help Red Hat users innovate and achieve their goals with our products and services with content they can trust.

Making open source more inclusive

Red Hat is committed to replacing problematic language in our code, documentation, and web properties. For more details, see the Red Hat Blog.

About Red Hat

We deliver hardened solutions that make it easier for enterprises to work across platforms and environments, from the core datacenter to the network edge.

© 2024 Red Hat, Inc.