Jump to: navigation, search

Cosmos Architecture Meeting 01-Nov-07

Attendees

  • Bill Muldoon
  • Saraub Dravid
  • David Whiteman
  • Balan Subramanian
  • Valentina Popescu
  • Jack Devine
  • Jimmy Mohsin
  • Hubert Leung
  • Don Ebright
  • Ali Mehregani
  • Mao Sheng
  • John Todd

Minutes

  • Localization
  • Namespaces (how do we choose this)
  • Lay the groundwork for serviceability

General Error Handling and Logging across COSMOS components

Logging Requirements

  • We would like to have consistency from the log to the component to the code
    • We should also be consistent in identification, including the instance identifier
    • Tie the instance identifier to the management endpoint (WSDM) that can be used to control the component
  • What is the structure of the logs--a physical view?
  • Need to define the "logging architecture"
  • Tracing/Logging levels
  • Need to have description of error log.
  • Would like to have one logical view of the log.
  • Standardized log format
    • Do we choose CBE or WEF
  • Symptoms

Error/Exceptions Requirement

  • Must log these
  • Examples of design documents that can be used as templates:

http://wiki.eclipse.org/index.php?title=COSMOS_Design_200222
http://wiki.eclipse.org/index.php?title=COSMOS_Design_204959

Next Steps


  • Prepare review & demo of existing tooling
    • Valentina to show/walk through TPTP in Detroit


  • Start identifying the "low hanging fruit" for the logging architecture (Jagmit w/Valentina's help)
  • First step of implementation would be to take all the "noise" currently going to the console to the logs (target i8)