Etienne Urbah proposed this requirement on 2010-03-17

Etienne Urbah's position on 2010-04-01

• Grid computing means distributed resources and frequent remote requests
• Therefore, a lot of things may go wrong, and experience proves that they will go wrong
• On errors, Grid middleware should ease :
  - Detection that an error has occurred   (we should NOT allow errors to propagate unnoticed)
  - Analysis of the error, leading to efficient error management (for example quickly resubmit job with corrected JSDL), and if necessary efficient software fix
• Therefore, each client or service should systematically detect all error cases
• Beside adequate logs useful post-mortem (see NF4) :  Each time an error occurs when processing a precise request, each client or service should immediately report the error details (component, version, configuration, credentials, input data, ...) to the submitter of the request
• So I vote FOR this requirement

Morris, Balazs and Etienne on 2010-04-27

• Source = General
• Spreadsheet ID = 148

Amsterdam meeting on 2010-04-28

• Agreed YES