wiki:SprintPlanningDHCP20111221

DHCP Sprint Planning Meeting: 2011-12-21

Attendees

Shane
Stephen
Tomasz

Actions from last Meeting

  • Stephen to draft some thoughts on defining options outside Kea
    In progress: The option definition document has been written but only covers IPv4. Feedback is wanted as are ideas for the IPv6 option definition.
  • Larissa to chase EOL of DHCPv3
    Outstanding.

Project Progress

Outstanding tasks from the task status spreadsheet (progress spreadsheet updated after this meeting attached) are listed below. Where are we on each one?

  • DHCPv4 state machine - respond to DISCOVER
  • DHCPv4 state machine - respond to REQUEST
    (Both tasks) Code was written today and used to configure the first client. Tests are being completed and will be checked in on Thursday. Stephen will review them as soon as they are checked in.
  • Hook definitions for server
    No feedback has been received for some time. The document as such will be the one submitted - we can revise it later, closer to the time we implement it. Stephen will close the ticket for that as part of this sprint.
  • DHCP benchmark tools - single packet exchange
  • DHCP benchmark tools - single packet exchange
    (Both tasks) Discussed below.
  • Option definition framework design - requirements
    The requirements document has been written and is being reviewed.
  • Option definition framework design - survey of methods and software
    This will be dropped.
  • Option definition framework design - high-level design
    The design document has been written but covers only IPv4 option definition. Feedback is needed as are suggestions for the IPv6 stuff.
  • Low-level network layer - interface detection
    The code currently does not meet the BIND 10 coding standards. It based on what Tomasz used in Dibbler which was in turn was new code loosely following the logic from another package. Owing to time constraints, it was agreed that this should be merged into master as-is, but that a ticket should be opened for refactoring.
  • Low-level network layer - transmission of data - V4
  • Low-level network layer - reception of data - V4
    Both of these tasks have been reviewed and ready to merge.

In addition, we added the task of:

  • Documentation
    Not started - the aim is to extend the adminstrator's guide.

Benchmarking

  • Benchmarking code (written in C) does not conform to BIND 10 standards - include and refactor?
    After some discussion, it was agreed that, like the interface detection code, it should be checked in as-is, and a ticket opened for refactoring. Stephen will merge it into the master branch on Thursday.
  • What do we do about all the benchmarking tickets
    If we are sure that tasks have been addressed, close them; otherwise, leave them for Francis to close.

Testing

We have the unit tests, really, what system testing should we do?

In the long-term, Lettuce-based system tests are needed. In the short term, we really need a manual system test (if nothing else, just two virtual machines, one running perfdhcp and the other running the server) just to make sure that it doesn't fall over immediately.

Shane will do some testing in the week after Christmas.

Outstanding Work

What to we still have to do to deliver the software?

  • Daemon needs to start as part of BIND 10
    As there is no msgq intergration as yet, so there is no point in trying to get it integrated into the BIND 10 framework now, especially as it does not serve real leases.
  • Still writes to stdout, needs to use the logging framework.

Refactoring tasks will be scheduled for the new year to correct these.

Documentation

As noted above, documentation will be included in the BIND 10 manual. To avoid suggesting that this is part of the "production" system, the documentation will conspicuously highlight the fact that this is a "technology preview".

Next Steps

  • Tomek - merge in current tickets as and when they are completed
  • Stephen - review DISCOVER and REQUEST code when completed tomorrow
  • Stephen - merge in perfdhcp and open ticket for refactoring
  • Tomek - open ticket for refactoring interface detection code and BIND 10 integration.
  • Tomek - documentation (after Xmas)
  • Shane - When Tomek has merge code, you will conduct sanity check test of the code.

Although Stephen is on holiday next week, he will be online at times. We will aim to construct the tarball next week and notify Comcast of its availability.

Last modified 6 years ago Last modified on Jan 26, 2012, 11:15:31 AM

Attachments (1)

Download all attachments as: .zip