kubernetes pod metrics in elasticsearch

Dashboard

cpu and memory of kubernetes pods collected by heapster using elasticsearch as a sink
Last updated: 10 months ago

Downloads: 91

  • Bildschirmfoto 2018-01-31 um 17.02.58.png
    Bildschirmfoto 2018-01-31 um 17.02.58.png
  • Bildschirmfoto 2018-01-31 um 17.09.36.png
    Bildschirmfoto 2018-01-31 um 17.09.36.png
  • Bildschirmfoto 2018-01-31 um 17.03.45.png
    Bildschirmfoto 2018-01-31 um 17.03.45.png

This shows cpu and memory metrics from kubernetes pods collected by heapster using elasticsearch as a sink. Cluster, namespace and pod are provided as template variables. Configure heapster to start with

/heapster
      --source=kubernetes:https://kubernetes.default
      --sink=elasticsearch:http://elasticsearch-host:9200?cluster_name=your_cluster_name

Configure two grafana data sources (for the same elasticsearch url), one using CpuMetricsTimestamp as a time field, another using MemoryMetricsTimestamp as a time field. Seems weird to have the same url in two data sources, but I couldn't get around it. See screenshot for index template. Set group by time interval in both data sources to your heapster --metric-resolution argument. Heapster default is 1m, so set group by time interval to >1m, see screenshot.

Software versions used:

  • Heapster version: v1.4.2
  • Elasticsearch version: 5.5.1
  • Grafana version: 4.4.3