Application Real-Time Monitoring Service:End of maintenance for cloud service self-monitoring metrics prefixed with arms_cms_ in Managed Service for Prometheus

Managed Service for Prometheus will end maintenance for the metrics prefixed with arms_cms_ in its cloud service self-monitoring collection pipeline. The change takes effect at 00:00:00 on October 9, 2026 (UTC+8). To keep your cloud service monitoring data flowing, migrate your cloud service monitoring metrics to the Integration Center of Cloud Monitor 2.0 before the effective time, and update your PromQL statements and alert rules.

The cloud service self-monitoring pipeline used a Cloud Monitor exporter that periodically called the Cloud Monitor DescribeMetric* operations to pull cloud service monitoring data. The exporter wrote the data to your Prometheus instances as metrics prefixed with arms_cms_. Because of limitations in its implementation, the pipeline could not guarantee collection stability or data continuity, and its issues were difficult to troubleshoot.

Determine whether you are affected

Run the following PromQL statement in your Prometheus instance:

count({__name__=~"arms_cms_.*"})

If the query returns data, your Prometheus instance contains metrics prefixed with arms_cms_. These metrics are written by the cloud service self-monitoring pipeline and are within the scope of this change.

These metrics typically appear in instances prefixes with [Cloud-Product-Prometheus-]. The platform automatically writes these metrics by calling the Cloud Monitor API. Other metrics written to the same Prometheus instance using Remote Write, managed agents, or custom collection tasks are not affected by this change.

Changes and impacts

• Effective time: 00:00:00 on October 9, 2026 (UTC+8).

• Discontinuation scope: Collection, writing, and support for issues related to metrics with the arms_cms_ prefix.

• Impacts: After the effective time, the current metrics will remain available. However, this pipeline will no longer receive feature updates, stability improvements, or bug fixes. Issues such as collection interruptions, missing metrics, and data latency will no longer be addressed.

Migrate to the Integration Center of Cloud Monitor 2.0

The Integration Center of Cloud Monitor 2.0 is where you integrate cloud service monitoring data. Compared with the legacy self-monitoring pipeline, it offers the following advantages:

• The data collection pipeline is centrally managed by the cloud service.

• You can view the integration status, collection tasks, and data health in the console to immediately locate any abnormalities.

• Metric metadata is centrally managed and you can select the integration scope by cloud service, instance, or workspace.

• Integration records can be queried, managed, and deleted.

The migration steps are as follows:

1. Log on to the Cloud Monitor 2.0 console. Select the target workspace and then navigate to Integration Center.

2. Click the Batch Integration of Cloud Service Metrics card, select an instance scope, and then click Next. Then, select the Alibaba Cloud services that you want to integrate and complete the configuration.

3. Return to the Integration Center page. Confirm that the integration is normal and that data is being collected for the new cloud service metrics.

4. While both pipelines run in parallel, verify that the newly integrated cloud service metrics are consistent with the metrics prefixed with arms_cms_ by referring to the cloud service metrics list. Then, update the PromQL statements in your monitoring dashboards and alert rules.

   Important

   The metric names, prefixes, and labels for metrics integrated through Cloud Monitor 2.0 differ from those of legacy metrics that use the arms_cms_ prefix. Therefore, you cannot reuse your existing PromQL statements. Rewrite the statements before the change takes effect, and migrate only after you confirm that the data is consistent.

Get help

If you encounter issues when you identify affected instances or migrate to Cloud Monitor 2.0, contact Alibaba Cloud Online Service in the management console or your account manager for assistance.