Runner upgrade

Purpose of the upgrade

The Alibaba Cloud DevOps Flow private build cluster architecture has been upgraded, replacing the legacy Agent with the new Runner. This upgrade introduces cluster load balancing and enhances observability, disaster recovery, and concurrency control.

• After the upgrade, pipeline build tasks use load balancing to utilize build cluster resources more efficiently.

• After the upgrade, the system switches to a new UI for private build clusters. This UI provides a clearer, card-based display of your clusters and simplifies onboarding new nodes. For more information, see Build Cluster.

Impact of the upgrade

• The upgrade applies to all machines in your organization's private build clusters.

• During the upgrade, a resident process for the Alibaba Cloud DevOps Runner will be installed and started on each build machine. This process exclusively pulls and runs build tasks and collects load data from the executor.

• The upgrade process does not affect any running pipeline tasks. The system will switch to the new Runner execution method only after the upgrade for the entire organization is complete.

Upgrade process

• During the upgrade, a resident process for the Alibaba Cloud DevOps Runner is installed and started on all machines in your organization's private build clusters.

• Confirm the upgrade result:

  • All successful: If the Runner is installed successfully on all machines in the build clusters, the organization-wide upgrade is successful. Pipelines will automatically switch to the new Runner execution method.

  • Partially or entirely failed: If the Runner installation fails on some or all machines, the organization-wide upgrade fails. Pipelines will continue to use the old Agent execution method, and running pipelines are not affected. Check the upgrade task details, and based on the failure messages, either retry the upgrade or skip the failed machines.

Step 1: Review the upgrade notice

Go to Flow > Global Settings > Build Cluster Management. Click Upgrade Runner to open the Comprehensive Build Runner Upgrade dialog box, and review the upgrade purpose, impact, and process.

The Upgrade Runner button is in the blue notification banner at the top of the page, which reads, "The new upgrade supports cluster load balancing. Upgrade your Runner to use this feature."

Step 2: Submit the upgrade

After you review the impact of the upgrade, click Upgrade Now to submit the upgrade task, which will run in the background. Do not modify the build cluster until the upgrade is complete. Alternatively, click Schedule Upgrade. The upgrade task will then run between 02:00 and 06:00 the next day. You can check the results the following day.

The dialog box displays the upgrade information: This upgrade replaces the Agent with the new Runner, enabling cluster load balancing, observability, disaster recovery, and concurrency control. The scope includes all machines in your private build clusters, including Alibaba Cloud ECS instances and on-premises hosts. During the upgrade, a resident process for the Alibaba Cloud DevOps Runner will be installed and started, which will not affect any running pipeline tasks.

Step 3: Check results and use new features

The upgrade process takes a few minutes. Refresh the page to view the result. After a successful upgrade:

• The Build Cluster Management page automatically switches to the new UI.

• Pipeline tasks will automatically use the new Runner to dispatch build tasks.

• You can add Windows machines as build nodes. When running a pipeline, you can specify a Windows build node to execute build tasks in the Default VM Environment.

Step 4: Handle failures

• If the upgrade fails, click Upgrade Failure Details. A dialog box appears showing the number of successful and failed machine upgrades, along with a list of failed machines and the failure reasons.

  Based on the failure messages, you can choose to Retry Upgrade or Skip Failed Machines.

• Resolve the issues based on the failure reasons (for reference, see Host Deployment Issues), then retry the upgrade. If a machine is no longer in use or can be handled later, you can click Skip for that machine.

  In the dialog box, you can view the failure reason for each instance, such as "Machine is offline. Please check machine status." After resolving the issue, if you need to manually mark the upgrade as successful, select the corresponding instance and click Set Runner Upgrade as Successful.

• When the upgrade status for all machines is either Succeeded or Skipped, you can mark the Runner upgrade for the entire organization as successful. The build cluster then switches to the new UI, enabling new features and automatically using the new Runner method for pipeline tasks.

Purpose of the upgrade

Alibaba Cloud DevOps Flow host group management has been upgraded, replacing the legacy Agent with the new Runner. This upgrade adds support for deployment to ARM hosts and streamlines node onboarding and improves observability.

Impact of the upgrade

• The upgrade applies to all machines in your organization's Flow host groups.

• During the upgrade, a resident process for the Alibaba Cloud DevOps Runner will be installed and started on each machine. This process exclusively pulls and runs deployment tasks and collects load data from the executor.

• The upgrade process does not affect any running pipeline tasks. The system will switch to the new Runner deployment method only after the upgrade for the entire organization is complete.

Upgrade process

• After you click Upgrade Now, the upgrade task starts immediately.

• During the upgrade, a resident process for the Alibaba Cloud DevOps Runner is installed and started on all hosts in your organization. Do not create or modify Flow host groups during the upgrade.

• Confirm the upgrade result:

  • All successful: If the Runner is installed successfully on all hosts, the organization-wide upgrade is successful. Pipeline host deployment tasks will automatically switch to the new Runner execution method.

  • Partially or entirely failed: If the Runner installation fails on some or all machines in a host group, the organization-wide upgrade fails. Host deployment tasks continue to use the old Agent method and are not affected. Check the upgrade task details, and based on the failure messages, either retry the upgrade or skip the failed machines.

Step 1: Review the upgrade notice

Go to Flow > Organization Settings > Host Group Management. Click Upgrade Runner to open the Host Group Runner Upgrade dialog box, and review the upgrade purpose, impact, and process.

The Upgrade Runner button is located in the blue notification banner at the top of the page.

Step 2: Submit the upgrade

After you review the impact of the upgrade, click Upgrade Now to submit the upgrade task, which will run in the background. Do not modify host groups until the upgrade is complete.

Step 3: Check results and use new features

The upgrade process takes a few minutes. Refresh the page to view the result. After a successful upgrade:

• The Flow Host Group Management page automatically switches to the new UI.

• The new Runner will automatically dispatch pipeline host deployment tasks.

• You can add Linux ARM hosts as environment deployment hosts.

Step 4: Handle failures

• If the upgrade fails, click Upgrade Failure Details. A dialog box appears showing the number of successful and failed machine upgrades, along with a list of failed machines and the failure reasons.

  Based on the failure messages, you can either retry the upgrade or skip the failed machines.

• Resolve the issues based on the failure reasons (for reference, see Host Deployment Issues), then retry the upgrade. If a machine is no longer in use or can be handled later, you can click Skip for that machine.

  The dialog box shows the specific failure reason for each machine, such as "Machine is offline. Please check machine status." You can select the corresponding machines and click Set Runner Upgrade as Successful to mark them as successful in bulk.

• When the upgrade status for all machines is either Succeeded or Skipped, you can mark the Runner upgrade for the entire organization as successful. The host group management page then switches to the new UI, which enables the new host group management features.