Network Monitoring

DVP can perform network monitoring between all cluster nodes, as well as between cluster nodes and external hosts. When monitoring is configured, each node sends ICMP packets twice per second to all other cluster nodes (and to optional external nodes) and exports data to the monitoring system.

The monitoring-ping module tracks any changes in the .status.addresses field of a node. If changes are detected, a hook is triggered that collects a complete list of node names and their addresses, and passes it to the DaemonSet, which recreates the pods. Thus, ping always checks an up-to-date list of nodes.

The monitoring-ping module must be enabled.

Adding Additional IP Addresses for Monitoring

To add additional monitoring IP addresses, use the externalTargets parameter of the monitoring-ping module.

Example module configuration:

apiVersion: deckhouse.io/v1alpha1
kind: ModuleConfig
metadata:
  name: monitoring-ping
spec:
  version: 1
  enabled: true
  settings:
    externalTargets:
    - name: google-primary
      host: 8.8.8.8
    - name: yaru
      host: ya.ru
    - host: youtube.com

The name field is used in Grafana to display related data. If the name field is not specified, the required host field is used.

Cluster Node Monitoring

To enable cluster node monitoring, you need to enable the monitoring-kubernetes module if it is not already enabled. You can enable cluster monitoring in the Deckhouse web interface, or using the following command:

d8 system module enable monitoring-kubernetes

Similarly, you can enable the monitoring-kubernetes-control-plane and extended-monitoring modules.