wiki:SprintPlanningDHCP20121129

BIND 10 DHCP Sprint Planning Call

29 November 2012, ~09:45 PDT/17:45 BST/18:45 CEST

Attendees

Jeff
Larissa
Marcin
Shawn
Stephen
Tomasz

Review of Past Sprint

Closed half the tickets in the sprint. Marcin is in the middle of #2491. Stephen had completed the MySQL backend (#2404) which is now in review.

Tasks Outstanding for Delivery

For options, these are:

  • Option definitions dhcp6, need to do dhcp4
  • Configuration parser for dhcp4
  • Option spaces
  • Setting up option definitions from the configuration manager

For the allocation engine, the outstanding tickets are listed in #2141 and are #2130, #2131, #2132, #2325, #2326 and #2327. Tomasz is working on #2325 and is cleaning up the code ready for review next week. One task that is needed is the implementation of expiration - we need to agree on a common approach.

Ticket Selection

It was agreed that #1609 and #2317 in the current sprint would be dropped - none of them have been started yet.

Tomasz asked that

  • #2325 (V6 renewal)
  • #2326 (V6 release)
  • #2486 (DHCP not receiving packets on CentOS)

Marcin requested new tickets be created for:

  • Create definitions for V4 options.
  • Configuration parser for V4 options
  • Generation of V4/V6 option spaces

Stephen added some minor tasks to tidy up some issues that had been encountered in reviews and builds:

  • #2396 (add toBytes to IOAddress)
  • #2479 (perfdhcp tests reference wrong directory)
  • #2513 (Move file from dhcp to dhcpsrv directory)
  • New ticket to add logging to dhcpsrv files

(regarding the last ticket, the intention had been to make libdhcp++ a modular DHCP library with no dependencies on packages such as logging. As a result, all errors are signalled by exceptions. However, splitting the server-specific parts of the code into libdhcpsrv means that logging can be added to them without affecting libdhcp++'s status as a modular DHCP library.)

Other

It was agreed that:

  • At some point the exception hierarchy should be tidied up. At the moment, all are derived from "exception" - there should be intermediate levels. (E.g. the MySQL backend exceptions could be derived from something like "MySqlException" which is in turn derived from "exception". This would allow code to cache MySQL-specific exception whilst ignoring the rest.)
  • Some of the options code now uses the DNS "Name" class. This is pulling in libdns++ which will make it more awkward to split the DNS and DHCP builds. Marcin was asked to post something to the bind10-dev list about this.
Last modified 5 years ago Last modified on Nov 30, 2012, 11:59:10 AM