Robert Thurlow - PowerPoint PPT Presentation

About This Presentation
Title:

Robert Thurlow

Description:

Minor rev via standards track RFC. Compound will include a rev number ... Quantized, intolerant of schedule slips. Thinking so far. Discussed at Connectathon 2003 ... – PowerPoint PPT presentation

Number of Views:19
Avg rating:3.0/5.0
Slides: 9
Provided by: ietf
Learn more at: https://www.ietf.org
Category:

less

Transcript and Presenter's Notes

Title: Robert Thurlow


1
NFSv4 Minor Revisions
Robert Thurlow Staff Engineer Solaris Data
Technologies
2
NFSv4 spec sets rules basics
  • Minor rev via standards track RFC
  • Compound will include a rev number
  • Error NFS4ERR-MINOR-VERS-MISMATCH
  • Negotiation via reserved Op 2 when we need it

3
NFSv4 spec sets rules don'ts
  • No procedures added or removed
  • No changes to args or results of ops
  • No changes to existing attrs
  • No deletion of ops/attrs/bits/enums
  • No new feature can be mandatory!
  • No mixing returned objects between compounds
    w/different minor rev

4
NFSv4 spec sets rules do's
  • Compound ops can be added
  • New attributes can be added
  • New mode bits can be added
  • Can add new enums or bits
  • Can declare op/attr as mandatory to not
    implement!
  • Can downgrade or upgrade features along
    MUST/SHOULD/MAY axis

5
What's unspecified?
  • Policy!
  • How to decide we need a new minor rev
  • How to schedule and plan a minor rev - enough
    stuff vs. train models
  • How to decide what's in and what's out
  • How to decide when we bump major revs
  • gt WG chairs own this
  • Negotiation and compatability

6
Possibilities
  • Enough stuff
  • Asynchronous, make judgement calls about when to
    revise
  • Not predictable -(
  • Features
  • Slot interesting features into revisions
  • Some features may have to wait unfairly
  • Train model
  • Revisions happen on schedule
  • ... as long as we have passengers
  • Quantized, intolerant of schedule slips

7
Thinking so far
  • Discussed at Connectathon 2003
  • Train model most attractive
  • We test at Connectathon every year!
  • Most predictable option
  • Need a formal schedule initial draft by X, solid
    draft by Y, doc ed combines by Z, test in March
  • Probably need appeals procedure via WG chairs
  • Need to change our working style spec
    prototype in spring, serious coding in winter

8
Robert Thurlow robert.thurlow_at_sun.com
Write a Comment
User Comments (0)
About PowerShow.com