The prometheus-agent add-on is the monitoring add-on of the Volcengine Managed Service for Prometheus (VMP), which collects the monitoring metrics data of Kubernetes clusters for the Volcano Engine VMP. This topic mainly introduces the change history of the prometheus-agent add-on. For more introductions and usage methods related to the add-on, refer to Add-on management.

March 2025

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.10.0-vke.1.24	v1.24 ≤ Cluster version ≤ v1.30	Maintained	Supports selecting DiscoveryRole. Supports adding global Labels for collected metrics. Enables Add-on scaling by default. Adds a permission configuration toggle for Secret metrics in kube-state-metrics.	For details, refer to Installing add-on.
v2.10.0-vke.1.20	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports selecting DiscoveryRole. Supports adding global Labels for collected metrics. Enables Add-on scaling by default. Adds a permission configuration toggle for Secret metrics in kube-state-metrics.	For details, refer to Installing add-ons.

November 2024

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.9.0-vke.1.24	v1.24 ≤ Cluster version ≤ v1.30	Maintained	Supports Kubernetes v1.30. Supports atomic writing of histograms. ServiceMonitor uses role: endpointslices by default. additional-monitoring-config supports more parameters.	ServiceMonitor uses role: endpointslices by default, which will cause the relabel configuration that previously used the meta endpoint label to fail.
v2.9.0-vke.1.20	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports atomic writing of histograms. ServiceMonitor uses role: endpointslices by default. additional-monitoring-config supports more parameters.	ServiceMonitor uses role: endpointslices by default, which causes the relabel configuration that previously used the meta endpoint label to fail.

June 2024

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.8.0-vke.1.24	v1.24 ≤ Cluster version ≤ v1.28	Maintained	Fixes the issue of duplicate collection of DCP register nodes. Adds the collection of self-monitoring metrics of the add-on.	None
v2.8.0-vke.1.20	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Fixes the issue of duplicate collection of DCP register nodes. Adds the collection of self-monitoring metrics of the add-on.	None

May 2024

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.7.0-vke.1.24	v1.24 ≤ Cluster version ≤ v1.28	Maintained	Supports VKEObservabilityRole.	None
v2.7.0-vke.1.20	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports VKEObservabilityRole.	None

March 2024

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.6.0-vke.1.24	v1.24 ≤ Cluster version ≤ v1.28	Maintained	Supports configuration of ExternalLabels.	Using the ExternalLabels feature will double the active time series in the corresponding workspace. It will recover after one hour. Check whether the workspace volume meets the requirements in advance. For details, refer to Custom metrics label.
v2.6.0-vke.1.20	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports configuration of ExternalLabels.	Using the ExternalLabels feature will double the active time series in the corresponding workspace. It will recover after one hour. Check if your workspace volume meets requirements in advance. For details, refer to Custom metrics label.
v2.5.0-vke.1.24	v1.24 ≤ Cluster version ≤ v1.28	Maintained	Supports Kubernetes v1.28. Upgrade the kube-state-metrics add-on, which supports more metrics and dimensions.	When upgrading the add-on, dashboard data of some metrics may be interrupted for 1 to 2 minutes. After upgrading the add-on, the active time series of the cluster's workspace will increase by approximately 10% for a short period. It will recover after one hour. Check in advance whether the maximum volume of active time series in the workspace is sufficient. The kube__annotations and kube__labels metrics in the kube-state-metrics add-on are no longer exposed, and therefore cannot be queried. For details, refer to Cannot query kube_namespace_label metrics. The kube_verticalpodautoscaler_.* metrics have been deprecated.
v2.5.0-vke.1.20	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Upgrade the kube-state-metrics add-on, which supports more metrics and metric dimensions.	When upgrading the add-on, some metrics will stop update for 1-2 minutes. After upgrading the add-on, the active time series of the workspace associated with clusters will temporarily increase by about 10%. It will return to the original level after one hour.Check in advance if the workspace's maximum capacity for active time series is sufficient.

February 2024

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.4.0	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Prohibit scheduling of DCP register nodes.	None

October 2023

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.3.0	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports writing the collected metrics in external Prometheus.	None
v2.2.0	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports configuration of metrics collection rules and deprecation of unnecessary metrics. The "instance label" has been removed from the kube-state-metrics metrics. A new RDMA collection rule has been added.	After upgrading the add-on to this version, the re-installation of the add-on of earlier versions is not supported. If you have such requirements, submit a ticket.

September 2023

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.1.0	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Standardizes the naming of API fields in the add-on configuration.	None

August 2023

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.0.5	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Adds ai.* Labels for kube_pod_labels.	None

July 2023

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.0.4	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports collecting Nydus metrics.	None
v2.0.3	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports ECS/VCI Role authentication. Users do not need to set workspace authentication information on the add-on configuration interface. Supports cluster workspaces. Users do not need to set a workspace on the add-on configuration interface.	None

June 2023

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.0.2	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports Kubernetes v1.26.	None
v2.0.1	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports service discovery through Service/Pod annotations.	None

May 2023

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v2.0.0	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports prometheus-operators (such as ServiceMonitor) for collecting Kubernetes Custom Resource Definitions (CRDs). Improves production-readiness so that the add-on can collect metrics from ultra-large-scale clusters.	This is a breaking change. You must uninstall the add-on and reinstall it. For detailed operations, refer to Volcengine Kubernetes Engine (VKE) Prometheus-agent add-on upgrade.

February 2023

Versions	Compatibility with Kubernetes version	Version Status	Description	Impact
v1.0.6	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports the collection of DCGM metrics on the Volcengine Container Instance (VCI).	None

December 2022

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v1.0.5	v1.20 ≤ Cluster version ≤ v1.24	Maintained	The add-on architecture is optimized.	None

November 2022

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v1.0.4	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports topology spreading configuration.	None

October 2022

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v1.0.3	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports the collection of mGPU metrics.	None

September 2022

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v1.0.2	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Adds VKE cluster ID for metrics.	None

August 2022

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v1.0.1	v1.20 ≤ Cluster version ≤ v1.24	Maintained	Supports hot updates of collection configuration.	None

July 2022

Versions	Compatibility with Kubernetes version	Version status	Description	Impact
v1.0.0	v1.20 ≤ Cluster version ≤ v1.24	No longer maintained	Supports the basic collection of Prometheus monitoring metrics.	None

Volcengine Kubernetes Engine