Improved Monitoring and Management Tools

Improved Management Tools


  • Integrated configuration tool that allows system-wide or server-specific configuration
  • Elimination of config files - servers contact a management station which provides config
    • Files still an option if desired
  • Dynamic reconfiguration
    • How to deal with clients in progress when configuration changes ...


  • online fsck
    • check and log problems
    • spot fix problems
    • offline for big problems
  • back pointers to check metadata structure or reassemble when broken

Management Tools

  • add magic to data storage area to verify correct
    • modify ping to check this
  • tool to initialize data storage area independent of metadata area
  • tool to initialize metadata area independent of data storage area
  • tool to view/modify preallocated handles
    • home side - what handles are allocated
    • remote side - what handles are preallocated
  • tool to check for duplicate handles


  • remove things marked for deletion
  • create replicants when needed
  • migration to lower tiers
  • quota issues


  • Currently have a wide selection of counters we can retrieve
  • Integrate with "collect" SNMP tool - allows data to be manipulated with a variety of standard tools
  • Need to be able to monitor by user - want to see which users are currently issuing requests, how many requests, what data volume
  • Instrument client to gather round trip request time
  • Measure caches hit rate (acache, ncache, sidcache, ucache)
  • develop automated system for adjusting timeouts, etc. based on traffic to optimize

Back to OrangeFS projects page