Sprint Planning meeting for sprint ending 2012-03-20


  • Jeremy
  • Jinmei
  • Larissa
  • Kevin
  • Mukund
  • Jeremy
  • Shane
  • Jelte
  • Michal
  • Stephen
  • Aharen
  • Kambe

Past Sprint Summary

Number of tickets closed: 21 Estimated count of closed: 85 (!) Number of tickets left new: 1 (#324, under active discussion on -dev) Estimated count of new: 5 Number of tickets assigned/in review: 11 Estimated count of active: 34

Current open defects: 265 (Last sprint: 258)

Ticket munging update: (see #1683)

Past Sprint Goals

  • Final cleanup (AND LIVE TESTING) of NSEC3

almost done?

  • Release-related issues

Release was made. :)

  • Performance improvement work

Background work done, but actual performance improvements still left to do. #1601, #1607 not quite done (in review)

Past Sprint General Review

How did release go?

Jeremy: If an administrator reads our release notes, they see change log entries. This doesn't show the big picture. So I have to guess or ask others. Something we need to discuss later on.

Jeremy: Release went well, no big difficulties.

Shane: Feature demo? Something with NSEC3? To be discussed on jabber and/or e-mail.

Jinmei: Also running multiple processes.

Shane: Maybe more interesting.

Jelte: Yeah that's a good one.

Jeremy: At release time I deploy software on a few systems and catch things at the last minute. There's not much else I can do since things aren't committed until a day ahead a time. I started a server with both auth and resolver, and it did not use the auth configuration, and it reused ports for both. I did resolver and auth shutdown and it started working.

Jelte: We need more deployment scenarios in general. We need more users; it helps if you have more failure scenarios.

Jinmei: Any plans to use the actual hours recorded?

Jelte: Still nothing fully formed yet. We should have enough information to see if and what we can do with it. I'll write it down as an action item for me to see if this information can be useful now.

Jinmei: If you can quickly list tickets that needed a longer time (like 20 hours) that may be helpful.

Jelte: any opinions on releasing halfway through a sprint?

Shane: I'm not a fan.

Larissa: Seems to add confusion.

Jinmei: What was the problem?

Shane: we had to prioritize some tickets so they would be done at the start of a sprint

Jelte: since we don't do hardening sprints, we used the 1st week as a hardening phase

jreed: It was good we didn't do the release one week early as suggested so we can prove NSEC3. jelte's lettuce test addition identified issues.

Jelte: Should we have hardening sprints or something like that? Without them, I liked it (releasing halfway through a sprint).

Jeremy: I think that sometimes we do need a hardening sprint.

Shane: we don't have to release every 3 sprints. We could do 3 sprints + 1 hardening sprint and then a release.

Larissa: Not a bad idea. Also, should a hardening sprint be the same length as the other sprints?

Jelte: A very good topic for the face to face meeting.

[ Jelte officially closes sprint ]

Next Sprint Goals

Note: Last full sprint of Y3.

Primary Goal:

  • Auth performance

Secondary goal:

  • NSEC3 in database datasources

Possible goals:

Jinmei: some things are missing for NSEC3 in database datasouces

  • OpenBSD build issues

Jinmei: Without an OpenBSD machine this is almost impossible to reproduce

Shane: How many people use virtual machines?

Jeremy: I set up a lab machine and I can add user accounts.

Shane: Not necessary to have OpenBSD on this sprint, but eventually it would be good...

Jeremy: Not OpenBSD specific, but OpenBSD is perhaps more pedantic.

  • Goals

Jelte: Can we get all of these goals?

Jinmei: We can't get NSEC3 in this sprint. My gut feeling is even specifying NSEC3 as a partial goal is probably too much.

Jelte: I want to have the different kinds of data souce equal in terms of functionality.

Jinmei: That's a good goal. It will help accellerating development, so it should be done.

Jelte: can we switch to the new data source API now?

Jinmei: we don't have hotspot caching

Michal: no way to combine with static data source (VERSION.BIND and this) (Ticket #1215)

Jinmei: we should first concentrate on performance

Shane: I think we didn't have enough performance tickets...

Jinmei: we have some additional things, at least 6 completely new tickets, and some incomplete tickets in review queue

Shane: perhaps NSEC3 in database datasources as a secondary goal?

Jinmei: the OpenBSD tickets are independent of each other, so we can achieve some progress in this area

Shane: One possibilty is to pick defects from the OpenBSD tickets

Stephen: How close are we to the end of year goals?

Stephen: If we do support NSEC3 even without the new model, do we need it by the end of year?

Shane: Performance is the last thing we need.

Stephen: Any showstopping or awkward bugs that need addressing before Y3 end?

Jinmei: One big missing piece is to support loading data into in-memory from the database. Without that we cannot use xfrin from in-memory. Not part of actual requirements for year 3.

Next Sprint Tickets

From goals:

  • Auth performance

#1608 implement ZoneFinder::Context::getAdditional() for task data source auth performance 5

in memory data source

#1747 refactor auth::Query so it's reusable task b10-auth auth performance 4 #1748 define AbstractRRset::isSameKind() and implement the task libdns++ auth performance 3

default version

#1749 implement RBNodeRRset::isSameKind() task data source auth performance 3 #1750 "add ""at origin"" option in ZoneFinder::find() and task data source auth performance 4

support it in in-memory"

#1753 use object pool for in-memory finder contexts task data source auth performance 5


  • NSEC3 in database datasources

defer: #1577 implement ZoneFinder::findNSEC3 in database data source task data source NSEC3 6 (depends on related tasks?) add: #1579 (5) #1758 and #1760 (both 5: 10)


running total: 39

  • Switch to new datasource API

(defer) #1207 Enable the data source factory task data source 5

Jinmei: Maybe we should first introduce a preliminary (perhaps ad-hoc) version of the interface in this sprint? A generic kind of configuration. #1207 is too vague - without clarification it does not seem to be feasible to work on this.

Shane: Probably we need all the tasks in this area and consider that as a goal for a separate sprint.

  • OpenBSD problems

Jeremy: #1644 already in review, but maybe we need the task where we reorder all the headers? (#1644) Shane: Lets expand the scope of #1644 to include all reordering. Jeremy: #1644 is done, lets just commit it after rolling across the build farm

#1726, #1727, #1728 are all in proposed queue as defects... (19)

#1640 is in review queue, fixes Ubuntu might fix OpenBSD too....

Mukund: #1639 can be trivially fixed

Jelte: I don't want to shove work to other people, but this is DHCP stuff. ;)


#1741 (2) #1698 (5)

Other from proposed: #1321 (4)

Other general: #1093 (1742 is duplicate) review only #1541 (0) review only

#1688 (5)

Stats: #1751 (6)

total: 80...

Jinmei: propose allowing Jelte to add some more things

Jelte: We'll keep it to this for now, and if it looks like we'll run out we'll add more

CURRENT PROPOSED QUEUE (The tickets not mentioned above are put back into the backlog): #556 Request for feedback in ./configure enhancement build system 2 #1321 Allow sending notify manually enhancement xfrout 4 #1393 reduce overhead in python logging enhancement logging 4 #1541 script for listing obsolete (fully merged) branches enhancement build system 0 #1579 Update database ZoneFinder::find() for negative cases task data source 5

of NSEC3-signed zones

#1606 make check fails in ubuntu defect build system 3 #1688 duplicate RR suprression in auth::Query task b10-auth 5 #1698 log initialization causes real fiasco for MacOS 10.7 defect logging 5 #1703 should not shutdown if a component is later added and it fails defect Boss of BIND 5 #1716 use DEBUG for BIND10_LOST_SOCKET_CONSUMER and defect Boss of BIND 2


#1726 SessionTests? hang on OpenBSD defect Unclassified 7 #1727 IntervalTimerTest?.invalidArgumentToIntervalTimer failure defect Unclassified 7 #1728 python import and versioned modules defect build system 5 #1732 document in-memory data source use in the guide task documentation 3 #1741 Duplicate RESOLVER_SHUTDOWN message defect resolver 2 #1742 Have message compiler exit with an error if a enhancement message-library 3

duplicate message ID is detected

#1745 adding pid to auth (and resolver) log ID defect logging 2 #1746 "cleanup: remove ""temporary"" defect data source 3


#1751 stats doesn't separate or consolidate reports from defect statistics 6

multiple auth instances

#1752 Measure performance hit of per-message filtering task logging 4 #1754 small issues in LabelSequence? defect libdns++ 4

[Phone call ended at 16:02 UTC.]

Last modified 6 years ago Last modified on Mar 6, 2012, 5:01:28 PM