Activities:

• Setup PoC (for understanding Prometheus and operators)
  • Create deployment using https://github.com/coreos/prometheus-operator
  • Get ideas from kube-prometheus, but deploy using Helm charts.
  • Some good information is available here: https://sysdig.com/blog/kubernetes-monitoring-prometheus-operator-part3/
  • Use node-exporter as daemon set
  • Use M3DB for storing time series DB
  • via remote-write and remote-read, store/retrieve data to/from M3DB.
  • Ensure that as more nodes are added to the cluster, prometheus starts to scrape them with no intervention.
  • Check the node-exporter statistics are put in M3DB and retrieved using PromQL or via Grafana.
    • Deploy node-exporter as a daemon-set
    • Configuring the operator to scrape these via service monitor CRD objects.
• Add Helm charts for operators
  • Parameterization should be sufficient enough to instantiate multiple times.
  • Start with this: https://github.com/helm/charts/tree/master/stable/prometheus-operator and modify as necessary to ensure that multiple instances of prometheus operator can be instantiated at various cloud sites.
  • For M3DB operator deployment,  we can start with this - https://github.com/m3db/m3db-operator/tree/master/helm/m3db-operator
• Bring up using ONAP-K8S Plugin
• Ensure that whatever test used during PoC continue to work.
• Bring up another copy of Prometheus in a different name space.
• Ensure that both prometheus copies are getting the data.
• Additional functionality:
  • Ensure that using remote_write and remote_read, data can be written in OpenTSDB (setup that is being brought up Raja).
  • Ensure that OpenTSDB reachability data is programmed via operators.
• CollectD:
  • Deploy collectD as daemon set
  • Add Collectd-to-Promethus exporter to the package.
  • Ensure that data from collectD is shown in both M3DB and OpenTSDB.
    *