BIND 10 Sprint Planning Meeting - 2011-08-16



Review of past sprint - Organizational

Michal: No problems with bottlenecks, but there were some collisions.

Shane: Caused a lot of problems, or minor?

Michal: Minor, but expecting more. Sometimes takes half an hour.

Jinmei: Can do sequentially to avoid collision, but my understanding is that we would prefer this type collision to be as fast as possible. We may be able to minimize conflicts with better planning, but in practice I think it is quite difficult.

Review of past sprint - Technical

Jelte: Data source refactoring design did have a number of details that were not complete, so people solved the same problem in different ways. That did result in a lot of communication.

Release Planning

Shane has updated the year 3 plan:

This was always a very full release, and we slipped a lot. As discussed last sprint planning, suggest the main focus should be to get IXFR done by Y3-4. Shane will discuss with steering committee after sprint planning.

Possible goals for Y3-4:

  • IXFR-out (Jinmei: #1162 is design task, implementation probably not too difficult)
  • IXFR-in (Jinmei: doable)
  • DNSSEC without NSEC3 for in-memory (Jinmei: If we stick with BIND 9 design should not be complicated; merge of SQLite with in-memory may cause problems)
  • msgq replacement (Shane: pending face to face meeting)
  • socket creator (Michal: #801, still some issues about moving sockets around; est 4 medium-sized tasks or something like that)
  • Boss process flexibility (Shane: I think not too hard)
  • Ability to save/load configuration to/from file and standalone syntax checker
    • change management processes, checkin configuration, backout configs, etc.
  • Work on performance
    • Profiling refactored data sources (Jeremy: not automated, but straightforward; understanding output is not straightforward!)
    • Performance improvements (Jinmei: until we complete all major features, including DNSSEC support in-memory, it doesn't make much sense to measure) (Michal: May have gut feeling about bottleneck - using full classes for names in the red/black tree.) (Jinmei: missing support of hotspot cache, should include that in task planning someday)
    • Adopt NSD-style prototype?
  • multiple core support (initially for auth) (Shane: pending face to face meeting)
  • 'rndc reload [zone]' equivalent to reload zone from file (Jinmei: current version is not scalable, may also depend on attractive technologies like shared memory)
    • also need zone loader rewrite SQLite & in-memory

Jeremy: Also some system & interoperability testing was supposed to be done in July and now.

Jeremy: I didn't see the stats planning (for b10-resolver), but I think we need some stats counters for the resolver. I hope to deploy the public resolver soon and it would be nice to deploy stats for it. I think a few counters would be a trivial task.

Shane: Would like to see IXFR-in, IXFR-out, and boss process.

Jinmei: Configuration stuff may not be so difficult.

Jinmei: What about NSEC support, RRSIG support, and so on for the new data source model?

Shane: Time for sprint planning... (may address it)

Sprint Planning

What tasks do we carry forwards?

Current tasks in review are 45 points

Accepted/assigned tasks are 18+ points

New tasks are 41 points

That is 104+ points, when we normally get about 70 points in a sprint.

Jinmei: Drop RR type implementation. We seemed to focus too much on that.

Shane: Agreed.

What tasks do we add?

We already have 5 defects on current sprint.

Jinmei: We did not complete any bug fixes. So we should put more on this.

Jeremy: #1152 already has a fix.

Jinmei: #1101 is minor but affects user experience.

Michal: #1153 looks annoying.

Michal: #1174 is bug related to refactoring.

Jelte: Once #1067 is merged we need to refactor the refactor.

#1173 to #1179 extend data source to include NSEC and NSEC3

Shane: Tickets for IXFR?

Jinmei: #1179 in this sprint. #1176 to #1178 as prereq for IXFR.

Jinmei: Also think about IXFR implementation in the Python code, but probably for the next sprint.

Kambe: Also #1175 (already accepted).

Jinmei: Suggest #1150.

Any Other Business

Jeremy: Anybody have anything planned to be merged to master in next 20 hours? For this week's release? ...

Last modified 7 years ago Last modified on Aug 17, 2011, 2:42:50 PM