Meeting: Feb 26, 2016

Present: Parag Mhashilkar, Marco Mambelli, Kevin Retzke, Brian Bockelman

Topics to discuss

All the ideas below are not final and still in the evaluation phase

  • Revamp monitoring code
  • Get rid of unsupported code base like html, javascriptrrd
  • Make it easy for the VOs and operators to configure their own dashboard/webpages and also aggregate information from multiple sources
  • Port existing content from RRDs to graphite backend
  • Questions
    • Can we make graphite and garafana easily deployable and configurable?
    • How do we handle the security of who can update? Does it support X509?
    • Do we need individual servers or can we have central repositories well managed by groups like OSG, FNAL, etc


My crude notes

  • Condor gangliad is supposed to be a generic condor monitord. Information available in the classads is periodically pushed to the ganglia or other monitoring backend. Should be in condor 8.5.3 - 8.5.4
  • Instead of graphite, influxdb might be a better option. Investigate.