In response to this:

This PR adds 6 CR tests(2 per platform for AWS, GCP and Azure) for CPMS boot image updates to the MCO's extended test suite.

The first test verifies the "All" mode for ControlPlaneMachineSets. As the CPMS resource is a singleton, this mode efficiently enables boot image updates for the whole cluster. The test then verifies that an update takes place by backdating the boot image and examining the CPMS resource. If the MCO has updated it correctly, the test then proceeds to verify the scale-up process by deleting an existing control plane machine. This last step can take a while as while a new node does scale-up and join the cluster quickly, the CPMS operator only proceeds to scale-down the old node after that. I think this is to maintain etcd quorum. Draining and deleting the old control plane node at the end of the test takes a while - the whole test seemed to vary between 15-20 minutes depending on the platform.

The second test verifies "None" mode for ControlPlaneMachinesets. This test is a lot more simpler than the first; it just checks if the backdated boot image does get updated by the MCO. The test is considered successful if the backdated boot image is left untouched.

This PR is based on #5332, which should merge soon. Only the last commit is relevant to this PR.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

• OWNERS [djoshy]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

In response to this:

This PR adds 6 CR tests(2 per platform for AWS, GCP and Azure) for CPMS boot image updates to the MCO's extended test suite.

The first test verifies the "All" mode for ControlPlaneMachineSets. As the CPMS resource is a singleton, this mode efficiently enables boot image updates for the whole cluster. The test then verifies that an update takes place by backdating the boot image and examining the CPMS resource. If the MCO has updated it correctly, the test then proceeds to verify the scale-up process by deleting an existing control plane machine. This last step can take a while as while a new node does scale-up and join the cluster quickly, the CPMS operator only proceeds to scale-down the old node after that. I think this is to maintain etcd quorum. Draining and deleting the old control plane node at the end of the test takes a while - the whole test seemed to vary between 15-20 minutes depending on the platform.

The second test verifies "None" mode for ControlPlaneMachinesets. This test is a lot more simpler than the first; it just checks if the backdated boot image does get updated by the MCO. The test is considered successful if the backdated boot image is left untouched.

This PR should not need QE verification since it is not affecting functions of the MCO, only test code.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

In response to this:

/payload-job periodic-ci-openshift-machine-config-operator-release-4.21-periodics-e2e-azure-mco-disruptive-techpreview
/payload-job periodic-ci-openshift-machine-config-operator-release-4.21-periodics-e2e-aws-mco-disruptive-techpreview
/payload-job periodic-ci-openshift-machine-config-operator-release-4.21-periodics-e2e-gcp-mco-disruptive-techpreview

/verified bypass

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

In response to this:

This PR adds 6 CR tests(2 per platform for AWS, GCP and Azure) for CPMS boot image updates to the MCO's extended test suite.

The first test verifies the "All" mode for ControlPlaneMachineSets. As the CPMS resource is a singleton, this mode effectively enables control plane boot image updates for the whole cluster. The test then verifies that an update takes place by backdating the boot image and examining the CPMS resource. If the MCO has updated it correctly, the test then proceeds to verify the scale-up process by deleting an existing control plane machine. This last step can take a while as while a new node does scale-up and join the cluster quickly, the CPMS operator only proceeds to scale-down the old node after that. I think this is to maintain etcd quorum. Draining and deleting the old control plane node at the end of the test takes a while - the whole test seemed to vary between 15-20 minutes depending on the platform.

The second test verifies "None" mode for ControlPlaneMachinesets. This test is a lot more simpler than the first; it just checks if the backdated boot image does get updated by the MCO. The test is considered successful if the backdated boot image is left untouched.

This PR should not need QE verification since it is not affecting functions of the MCO, only test code.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

In response to this:

This PR adds 6 CR tests(2 per platform for AWS, GCP and Azure) for CPMS boot image updates to the MCO's extended test suite.

The first test verifies the "All" mode for ControlPlaneMachineSets. As the CPMS resource is a singleton, this mode effectively enables control plane boot image updates for the whole cluster. The test then checks that an update takes place by backdating the boot image and examining the CPMS resource. If the MCO has updated it correctly, the test then proceeds to verify the scale-up process by deleting an existing control plane machine. This last step can take a while as while a new node does scale-up and join the cluster quickly, the CPMS operator only proceeds to scale-down the old node after that. I think this is to maintain etcd quorum. Draining and deleting the old control plane node at the end of the test takes a while - the whole test seemed to vary between 15-20 minutes depending on the platform.

The second test verifies "None" mode for ControlPlaneMachinesets. This test is a lot more simpler than the first; it just checks if the backdated boot image does get updated by the MCO. The test is considered successful if the backdated boot image is left untouched.

This PR should not need QE verification since it is not affecting functions of the MCO, only test code.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

In response to this:

This PR adds 6 CR tests(2 per platform for AWS, GCP and Azure) for CPMS boot image updates to the MCO's extended test suite.

The first test verifies the "All" mode for ControlPlaneMachineSets. As the CPMS resource is a singleton, this mode effectively enables control plane boot image updates for the whole cluster. The test then checks that an update takes place by backdating the boot image and examining the CPMS resource. If the MCO has updated it correctly, the test then proceeds to verify the scale-up process by deleting an existing control plane machine. This last step can take a while as while a new node does scale-up and join the cluster quickly, the CPMS operator only proceeds to scale-down the old node after that. I think this is to maintain etcd quorum. Draining and deleting the old control plane node at the end of the test takes a while - the whole test seemed to vary between 15-20 minutes depending on the platform.

The second test verifies "None" mode for ControlPlaneMachinesets. This test is a lot more simpler than the first; it just checks if the backdated boot image does get updated by the MCO. The test is considered successful if the backdated boot image is left untouched.

This PR should not need QE verification since it is not affecting any functions of the MCO, it only adds new tests.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.