2018-04-03 Meeting notes: metrics demo




  • Demo of metrics offerings as they exist for RNode-0.2.1 to help make decisions about next steps for future releases

Discussion items


 Docker compose
  • Requires maintenance
  • Limits those working in more professional environments
  • More for a getting started audience
  • Requires testing with each release 

  • Node to run with Prometheus integration
  • When you run the node in the default mode, Prometheus needs to be able to scrape
    • This is in place
    • Next step
      • Expose extensive metrics on the system and JVM
      • Node to expose the metrics
  • Metrics to expose for Node-0.3
    • CPU - CPU seconds consumed, it's a counter, CPU seconds consumed/interval of scraping 
    • Ram
    • Disk 
    • Network core metrics at the node level
    • JVM performance
    • Garbage collection
    • Size of memory pools
    • Consumption of memory pools
    • Doc needs to describe how to integrate with an existing Prometheus, (if wanted) how to integrate with Docker compose
    • Doc how to export metrics from Scala
      • Pawel to create template 
      • Jeremy can then fill it out
  • Testing metrics
    • We need to create test to validate the outcomes
      • Node comes up
      • Node publishes metrics
      • Metrics are scrapable with Prometheus
      • Metrics are scrapable with Docker compose and match
  • At end
    • Rholang team should be able to integrate with the metric monad to look at performance 
    • Everyone needs to be able to find the Kamon interface and interact with it
    • Need a doc for this
    • Need a doc for how to capture metrics
      • Don't d
      • Publish counters and not gauges
      • Publish the numerator and denominator as separate metrics
  • IDEA get Pawel to write one metric in the way he likes as an example
    • Set up a session and record

Action items