Help Center/ Cloud Container Engine_Autopilot/ FAQs/ Chart and Add-on/ How Do I Adjust the Add-on Resource Quotas Based on the Number of Pods in a Cluster?
Updated on 2025-09-02 GMT+08:00
View PDF
Share

How Do I Adjust the Add-on Resource Quotas Based on the Number of Pods in a Cluster?

After the number of pods in a CCE Autopilot cluster is adjusted, you need to adjust the add-on resource quotas based on the number of pods to ensure that add-on instances can run normally. For example, if you change the number of pods from 50 to 200, you need to increase the vCPUs and memory of the add-on instances to prevent exceptions such as OOM caused by the scheduling of too many nodes. Only add-on resource quotas required by 50, 200, and 1,000 pods are provided in this FAQ. For the add-on resource quotas of other number of pods, you can convert them based on the values in the tables in this FAQ. If you are not sure about the add-on resource quotas, contact us for consultation by submitting a service ticket.

The values in the tables in this FAQ are recommended for typical service scenarios. You can also adjust the add-on resource specifications based on the service scenarios and monitoring data.

CoreDNS

When the number of pods in a CCE Autopilot cluster increases, the load on the component (CoreDNS) instances also increases. You need to adjust the number of component instances and their vCPU and memory quotas based on the number of pods. For details, see Table 1.

Table 1 Recommended resource quotas for CoreDNS

Number of Pods

Number of Component Instances

Component vCPU Limit

Component Memory Limit (MiB)

50

2

500m (0.5 vCPUs)

2,048

200

2

1,000m (1 vCPU)

2,048

1,000

2

2,000m (2 vCPUs)

4,096

CCE Advanced HPA

When the number of pods in a CCE Autopilot cluster increases, the load on the component (customedhpa-controller) instances also increases. You need to adjust the number of component instances and their vCPU and memory quotas based on the number of pods. For details, see Table 2.

Table 2 Recommended resource quotas for CCE Advanced HPA

Number of Pods

Number of Component Instances

Component vCPU Limit

Component Memory Limit (MiB)

50

2

1,000m (1 vCPU)

2,048

200

2

4,000m (4 vCPUs)

4,096

1,000

2

8,000m (8 vCPUs)

8,192

Cloud Native Cluster Monitoring

When the number of pods in a CCE Autopilot cluster increases, the load on the component instances of the add-on also increases. You need to adjust the number of collected shards based on the number of pods.

  • To enable Prometheus HA for Prometheus, refer to the following configuration.
    Table 3 Resource quotas for the Cloud Native Cluster Monitoring add-ons, with Prometheus HA enabled

    Number of Pods

    Number of Collected Shards

    Component

    Number of Component Instances

    Component vCPU Limit

    Component Memory Limit (MiB)

    50

    2

    prometheus-operator

    2

    1,000m (1 vCPU)

    2,048

    prometheus-lightweight-shard-0

    2

    2,000m (2 vCPUs)

    4,096

    prometheus-lightweight-shard-1

    2

    2,000m (2 vCPUs)

    4,096

    kube-state-metrics

    2

    1,000m (1 vCPU)

    2,048

    200

    10

    prometheus-operator

    2

    1,000m (1 vCPU)

    2,048

    prometheus-lightweight-shard-x

    (A set of the prometheus-lightweight components. The quota of each component is the same, and the number of components increases with the number of collected shards. x indicates the component sequence number.)

    2

    2,000m (2 vCPUs)

    4,096

    kube-state-metrics

    2

    1,000m (1 vCPU)

    2,048

    1,000

    50

    prometheus-operator

    2

    1,000m (1 vCPU)

    2,048

    prometheus-lightweight-shard-x

    2

    2,000m (2 vCPUs)

    4,096

    kube-state-metrics

    2

    1,000m (1 vCPU)

    2,048
  • If Prometheus HA is not enabled, refer to the following configuration.
    Table 4 Resource quotas for the Cloud Native Cluster Monitoring add-ons, with Prometheus HA disabled

    Number of Pods

    Number of Collected Shards

    Component

    Number of Component Instances

    Component vCPU Limit

    Component Memory Limit (MiB)

    50

    1

    prometheus-operator

    1

    1,000m (1 vCPU)

    2,048

    prometheus-lightweight

    1

    2,000m (2 vCPUs)

    4,096

    kube-state-metrics

    1

    1,000m (1 vCPU)

    2,048

    200

    5

    prometheus-operator

    1

    1,000m (1 vCPU)

    2,048

    prometheus-lightweight-shard-x

    (A set of the prometheus-lightweight components. The quota of each component is the same, and the number of components increases with the number of collected shards. x indicates the component sequence number.)

    1

    2,000m (2 vCPUs)

    4,096

    kube-state-metrics

    1

    1,000m (1 vCPU)

    2,048

    1,000

    25

    prometheus-operator

    1

    1,000m (1 vCPU)

    2,048

    prometheus-lightweight-shard-x

    1

    2,000m (2 vCPUs)

    4,096

    kube-state-metrics

    1

    1,000m (1 vCPU)

    2,048

Other Add-ons

Resource quotas of other add-ons may also be insufficient due to cluster scale expansion. If, for example, the CPU or memory usage of the add-on instances increases and even OOM occurs, modify the resource quotas as required. If you are not sure about the add-on resource quotas, contact us for consultation by submitting a service ticket.

Parent Topic: Chart and Add-on