BIND 10 Team Call



  • Mukund
  • Stephen
  • Kambe
  • Aharen
  • Shane
  • Jelte
  • Marcin
  • Tomek
  • Larissa
  • Fujiwara
  • jinmei
  • Jeremy
  • Michal
  • Jeff

Sprint Check-in (10 minutes)

Usual progress check of sprint

Jelte: In terms of tickets okay... but the problem with most systems failing right now Revert the merge? Mukund: No way to make it work on all these platforms... works for me
Shane: Can request builder make a branch
Jelte: That's inefficent (may take a day). Everyone should have access to the test systems.
Jeremy: Limited disk space on the systems though. :(
Jeremy: 3 systems where DDNS did not work
Jeremy: might be faster to run a VM on your own hardware in the long run
Shane: kind of a hassle
Jelte: I am doing it. They do go stale.
Michal: would be nice if we could download the disk image...

Mukund: Debian failing, Solaris failing, MacOS / CentOS have failure with pthread return value (patch exists), MacOS has issue linking on some systems

Jinmei: so will we revert it?
Mukund: if more than 12 hours or so...
Jinmei: if you can complete it within your day then fine, else revert
Mukund: my day is over! I'll try to discuss with Jeremy and fix in the morning
Jinmei: should set up account for Mukund to log into these machines

Shane: #2348?
Jinmei: I'm planning on this today or tomorrow
Shane: Awesome

Jinmei: Consider task prioritizing?
Jelte: I'll lower less important tasks.

Packaging (DHCP-initiated) (20 minutes)

Stephen thinks it will be useful to split BIND 10 into several separate packages. Now is the time! botan not used by dhcp; dhcp's mysql is not used by dns has code examples for selecting parts to build/install

Larissa: decoupling should happen before the beta
Michal: should have them reversed, "--enable-dhcp", "--enable-dns"
Stephen: makes us have to do ./configure --enable-dns instead of ./configure.
Jinmei: separation is a good thing. just create a task and assign it to someone

Stephen: what about versioning?
Stephen: logically we should have separate packages, bind10-common, bind10-dhcp, bind10-dns
Michal: not against separate source tarballs, but I think we still want 1 big repository

Jelte: higher priority to reviewing library layout?
Tomek: there are still some tools that don't use Python at all (perfdhcp)
Jelte: in the future we may have bind10-dnsutils, bind10-dhcputils... but for now not necessary!

Tomek: also consider how the build farm will work... test some combinations
Jinmei: I think we should do that. It may not be 100% complete (trying every possible combination), but at least we should check the major subsets, otherwise some combination will not work

Jinmei: core developers will still need to build everything on their machines. Suspect cannot avoid the situation of installing something that is not necessary for their environment.

Larissa: I will discuss with Joao.

Jeremy: not separate source tarballs, but building separately and installing separately
Stephen: need to install MySQL on target systems. ;)

Alpha/Beta? Planning (10 minutes)

Shane will go over the current expectations for Alpha/Beta? releases.
maybe meta ticket ?

Need to improve installation

alpha-2 2012-10-25, includes background loading ;)

Beta in December - zone loading rewrite

Joao wants alpha-3, mid-November?

Recursive Resolver, Hiring (10 minutes)

Shane discusses the expectations regarding developing the recursive resolver, and hopes of more hands to help out.

Trying to get a new developer in two weeks. And another in early 2013.

Are there job descriptions available? Some of us might know some qualified people.
(Shane will post link into chat room.)

Goal is to release recursive server in 2013. In January begin looking at resolver seriously. May be able to do (continue) some research prior to that.

(Shane to send Jinmei real data from Comcast.)

Sprint Planning Tweak Proposal (15 minutes)

Going to try it. :)

Reviews vs New Development

Michal: Feeling more reviews than development tasks. Is that me too?
Mukund: I've noticed more development work than reviewing, because when I go to take a ticket it is already being reviewed.
Shane: time zone?
Jinmei: I don't think so. Maybe you pick bigger tickets so they take longer to complete? Pick up relatively easier tickets, but don't know if it really helps...
Michal: I might try having a limit on how many review I take in a row. Sometimes I take 4 or 5 in a row.

Jinmei: I don't have the impression of more review than development. Not happening to me.
Jelte: Also not noticed significant changes. Have had a few more reviews due to statistics work, but that should not be significant.

Shane: I think capping reviews is reasonable. If you've done several and are ready for new development work, then I think it is okay to take a new ticket.

#include <this> or "this"
Michal: "quote" for local files, many people prefer to use <pointy-brackets> for everything
Jinmei: need to read full thread, consistency is good in general, not sure what the result will look like
Jinmei: gut feeling is to leave it flexible, but there may be bad things that we have to note and include in the guidelines


  • improve

Larissa: goal is to not have users go to, but have them go to the site
Shane: should have this before beta

  • security procedures

Jeff is now the one running ISC security procedures. Did a "dry run" the last time, will lean on Larissa for the next one. Shane will talk with Jeff about this.

  • DB backend use cases

"The use of MySQL is for legacy reasons. The customer has all DNS data in a MySQL database already, and by using a DNS server that also utilized MySQL the existing database and knowledge can be re-used. The motivation is to keep the existing provisioning infrastructure as stable as possible (because it has many interfaces to several backend systems and registrar APIs).

The idea is to have a MySQL based DNS Server where the DNS data (non DNSSEC) is managed in MySQL via an existing Web-UI and home-grown APIs, and the DNS Server will take care of the DNSSEC management (signing, signature refresh, key rollover)."

Jelte: implies a non-closed database


    We don't care as long as it's light and flexible
    It has to be what use already use
    I need it to be of a category of databases (SQL, noSQL, relational, ...)

A lot of conflicting use cases.

Jelte: nobody mentioned use case "no loading time"?
Larissa: that is another valid perspective
Larissa: can pull together what we already have

Mukund: with this use case, they want to use MySQL but don't care about the schema
Shane: we might be able to do things with foreign keys linking into existing schemas or views
Mukund: we could Unix ODBC libraries and support several databases
Shane: for open source I think we can just install, for proprietary... meh

Jinmei: without knowing real users real requirements, everything is a guess
Jinmei: we need to collect real input from real users and summarize that somewhere

Over at 16:00 UTC.

Last modified 6 years ago Last modified on Oct 26, 2012, 12:28:18 PM