wiki:SprintPlanning20120124

Sprint Planning Meeting for Sprint ending 2012-02-07


Held on 2012-01-24

Attendees


Jeremy
Michal
Aharen
Jelte
Fujiwara
Shane
Kambe
Larissa
Stephen (eventually)
Jinmei (after eventually)

Past Sprint


Past sprint was, due to the F2F, and the number of unknowns, a bit of a weird one; mostly exploratory tasks and build issues. It was also only 5 days, some of which a few people were still in another F2F or traveling.

Summary:

Tickets closed: 13 Est. of closed tickets: 27 (but lot of the tickets did not get estimates) Tickets in review: 3 Est. of tickets in review: 8 Tickets left unassigned: 2

Shane: Were non-estimated exploratory?
Jelte: Yes, and everything that came up in the face to face were not estimated.

http://tinyurl.com/7gmd7jw

Stephen: Will #1566 be split into two?
Jelte: I think #1566 was to define benchmarks, and then a different task would be to do them. Then we can actually measure our performance increases when we do them.
Stephen: #1566 does not actually describe what it is supposed to do...

Stephen: We've also got 2 unassigned tickets in review. #1568 is actually "on hold". The second one is #1448, which is SNMP. Are we looking for someone with SNMP experience for that?
Jelte: Yes I'm going to ask Shawn about that.

Jelte: Should we do exploratory sprints more often?
Stephen: Like looking at performance?
Jelte: Forcing us to do design work in advance. I was going to propose we do hardening sprints, perhaps this could be added as well.

Review:

Next Sprint


Goals:

  • Base for NSEC3 in-mem implementation
  • In-mem performance improvements based on F2F research

NSEC3 inmem 'base':

#1438     NSEC3: consistent selection of NSEC3PARAM     enhancement     5 
#1574     Add support for loading NSEC3 RRsets to in memory data source     task     7 
#1575     move data_source.cc:Nsec3Param to libdns++     task     4 
#1578     Update in memory ZoneFinder::find() for negative cases of NSEC3-signed zones     task     5 

Stephen: Jinmei envisages that we only support one NSEC3PARAM record and one NSEC3 chain. If this is true #1438 can be put into the background.
Jelte: But we should have a plan for this.
Michal: I also noticed that the tickets for multiple auth running are not included in any of these groups. Is that overlooked?
Jelte: I simply looked which tasks had "feature depending on" as "performance". So they are not excluded on purpose.
Michal: These are #1595 and #1596.

Jelte: Do we just want to put in memory stuff?
Shane: Perhaps we can make base stuff CRITICAL and the protocol stuff less priority than performance tasks, for example.
Jelte: We had talk of changing labels for priorities.
Shane: Seems reasonable.

21

#1570
#1571
#1573


NSEC3 inmem 'protocol':
#1580     auth::Query NSEC3 support: Name Error case     task     4
#1581     auth::Query NSEC3 support: No Data, non DS case     task     4
#1582     auth::Query NSEC3 support: No Data, DS case     task     4
#1584     auth::Query NSEC3 support: Wildcard answer case     task     5
#1585     auth::Query NSEC3 support: Unsigned referrals case     task     4
#1586     Check for queries for NSEC3 owner names in in-memory data source     task     4 

25

Performance:


#1601     pass reference instead of shared pointer within AuthSrv     auth performance     task     3
#1604     reorganize RRset class hierarchy to allow further optimization     auth performance     task     5
#1605     introduce special RRset for in memory data source     auth performance     task     5
#1607     define ZoneFinder::findAdditional() and implement its default version     auth performance     task     4 
#1595 Make the share name configurable                                         3
#1596 Addressing of multiple instances of the same component         5

Jelte: I think a lot of the performance work may depend on #1604.
Shane: So, suggest that be the highest priority non-NSEC3 task.

Jelte: I would like to do #1595 and #1596 as well, as they may cost more work.

NSEC3 base:                21
#1604, #1595, #1596:    13

Shane: Suggest then defects then NSEC3 inmem.
Larissa: Yeah sometimes we don't get a lot past 40 points.

25

Other proposed tickets:

Defects:

#640
#1542    4
#1543    3

Shane: #1551?
Jelte: Depends on #1614...

{{ #1614 5 #1551 4

Running total: 50 + #640 }}}

Jelte: Should we get through that we can start on the other NSEC3.
Shane: Yup.

Statistics?
#1448 document basic design of stats snmp interface (continued)
#1400 auth per qtype statistics (continued)
#1613 auth per rcode statistics (new)

Jelte: Just add #1613?
Shane: Yes.
Jelte: I think this will be quite similar to one that is done/being worked on.

Jeremy: What about lettuce tests?
Shane: Do we have any tickets for this?
Stephen: Wouldn't it make sense to add what we need first?
Shane: We did have a regression...
Stephen: We should avoid the temptation to extend tests because that's what we said we would do, and then miss the deadline.
Jeremy: We could ask Cathy for this.
Jelte: When are they back?
Larissa: Next week.
Jelte: I propose that next week we talk to Cathy about that.
Stephen: I would suggest if Cathy is doing tests, we should do NSEC3 tests.
Shane: Seems reasonable to me.

Michal: What about tickets that are already started? Do we carry them over?
Jelte: I usually move everything over that is not in state "new".

Shane: Ticket #1400 stuck?
Aharen: Just created branch for it, but almost no other work. If no time to work on this in the next sprint, I'll make it unassigned.
Jelte: Probably we should not add #1613.

Jinmei: #1570, #1571, #1573 necessary for NSEC3.
Jinmei: DS tickets may also be considered 'protocol work', even though a number of NSEC3 tickets may depend on them. I'm not sure what's the best approach. If we are okay with having a relatively higher number of estimated points. The easiest way is just to include all 3 tickets.
Jelte: Then we won't add #1580 to #1586.
Jinmei: #1580 can start right now... there should be no dependency. Add #1570, #1571, #1573, remove #1581 and #1582 and #1585.
Jelte: Okay but then we're way over 80.
Jinmei: So then it's okay to remove them all.

Last modified 6 years ago Last modified on Feb 2, 2012, 1:39:05 PM