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
namefield is used in Grafana to display related data. If thenamefield is not specified, the requiredhostfield 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.