Configuring Prometheus and Grafana

The helm chart for PolarDB-X Monitor uses the default configurations for Prometheus and Grafana. If you wish to modify the related configurations, you can install or upgrade the PolarDB-X Monitor using the following commands and override the default settings with a values.yaml file.

helm install --namespace polardbx-monitor polardbx-monitor polardbx-monitor-1.6.0.tgz -f values.yaml

or:

helm upgrade --namespace polardbx-monitor polardbx-monitor polardbx-monitor-1.6.0.tgz -f values.yaml

The values.yaml file contains configuration items for Prometheus and Grafana. Below are configuration examples for common scenarios. For a detailed list of configurations, please visit: values.yaml 。

Configuring LoadBalancer

If your K8s cluster supports LoadBalancer, you can specify the following configuration when installing or upgrading PolarDB-X Monitor with the -f parameter:

monitors:
  grafana:
    serviceType: LoadBalancer
  prometheus:
    serviceType: LoadBalancer

Persisting Monitoring Data

By default, the monitoring data for the created Prometheus cluster is not persisted, which poses a risk of data loss. You can specify a directory for data persistence in your values.yaml file as follows:

monitors:
  prometheus:
    persist: true
    # Storage class supported within the K8s cluster
    storageClassName: ssd
    # Size of the storage space
    storageRequest: 100G

Configuring the Specifications for Prometheus and Grafana

In the default configuration, the Prometheus cluster includes one node with 8C16G of resources allocated, and Grafana includes one node with 4C8G of resources allocated. You can modify the specifications and the number of nodes for the Prometheus and Grafana clusters using the following configuration items:

monitors:
  grafana:
    resources:
      requests:
        cpu: 1000m
        memory: 2Gi
      limits:
        cpu: 2000m
        memory: 8Gi
  prometheus:
    resources:
      requests:
        cpu: 1000m
        memory: 2Gi
      limits:
        cpu: 2000m
        memory: 8Gi