OpenBSD Journal

N2K8 Hackathon Summary Part 1

Contributed by merdely on from the sushi-anyone? dept.

Mark Uemura (mtu@) shares the first of a several part series summarizing n2k8:

n2k8 group

Network Hackathon (Part 1) - May 5-10, 2008, Ito, Japan

I had the pleasure and privilege of attending one of the big hackathons in Calgary a couple years back. I remember Theo telling me that the hackathons, both big and small, were very important to the Project. He would go on to say that, "it is where magic happens". After participating at my first big hackathon, I totally understood what he meant. There is a communication and synergy that takes place at hackathons where you see the 'magic' just happen and you go, "Wow!".

Mark continues below.

I think that most OpenBSD users would agree that they are always pleasantly surprised come every release. All this happens with developers so geographically dispersed; it's amazing that anything actually gets done. It is a tribute to Theo's genius and forethought to make the CVS repository open to the public. OpenBSD was the first to do this.

This year's Network mini-Hackathon was held in a 68 year old traditional Japanese Ryokan (kind of like a hotel but Japanese style) with a natural Onsen (Hot Spring) inside the hotel in a fairly remote city called Ito about two hours outside of Tokyo. We chose the location as it was typically Japanese and far away from the myriad of distractions in Tokyo.

For many of the hackers, this was their first trip to Japan, if not Asia. We were a little concerned about the food especially having to cater to all the vegetarians that were coming. However, as it turns out, the Ryokan cooks prepared different meals (what seemed like eight course meals but pretty much served at the same time) every night and seemed to satisfy almost everyone. Comments such as "oh, there are eyes on the end of these noodles!" from Reyk were not uncommon. Actually comments started flying when the more exotic but traditional stuff arrived like the Natto (sticky, stringy and kind of smelly fermented soya beans). Regardless, everyone enjoyed the food especially in this unique setting.

The following OpenBSD Hackers were able to make it to the Network Hackathon:

mcbride@ (Ryan McBride), kjc@ (Kenjiro Cho), markus@ (Markus Friedl), mpf@ (Marco Pfatschbacher), henning@ (Henning Brauer), reyk@ (Reyk Floeter), deraadt@ (Theo de Raadt), pyr@ (Pierre-Yves Ritschard), claudio@ (Claudio Jeker), krw@ (Ken Westerback), beck@ (Bob Beck), djm@ (Damien Miller), dlg@ (David Gwynne) and norby@ (Esben Norby)

OpenBSD Testers: sakurai@ (Tomoyuku Sakurai) and mtu@ (Mark Uemura)

Japanese Developers invited by Kenjiro-san: Yojiro Uo, Keiichi Shima

Japanese/English Translator and cultural advisor: Ayako Uo

Bob Beck has always impressed me with with his humour, intelligence and colourful but informative posts besides all the useful things that he has done for the Project. Well, believe me when I say that he is just as colourful in person. There were a few times when we had to ask everyone to lower their voices but that was just our Japanese way of saying "Hey Bob, you're waking the other guests." That is how heated and interesting the discussions were from time to time.

I wanted to start this article off with one of the first big surprises of the hackathon. Almost right from the get-go, Bob pulls a rabbit from his hat with the DHCPD Synchronisation diff, now in -current, and here's how simple the setup is.

As an example, if you want to run your dhcp server on 192.168.1.2 and 192.168.1.3, after giving them the same /etc/dhcpd.conf(5) setup, you would do:

on .2:

dd if=/dev/arandom of=/var/db/dhcpd.key bs=2048 count=1
scp /var/db/dhcpd.key 192.168.1.3:/var/db/dhcpd.key
dhcpd -Y 192.168.1.3 -y 192.168.1.2 em0

on .3:

dhcpd -Y 192.168.1.2 -y 192.168.1.3 em0

Here's what Bob has to say about this and the work he did:

beck@ hacking "Well, let's see, I wrote and committed the changes to dhcpd(8) to allow for synchronisation of leases so you can run more than one. I worked a bit with Reyk on some forthcoming changes to dhclient(1) for link state detection, and got some good ideas at how to make our dhcp client much better at dealing with routes now that I had some good discussions with Claudio about the new route priority stuff he put in. By in large it's a lot of discussions that happened here that help me out a lot. While we got a lot of things into the tree this week, I think a lot more got started in a way such that they'll be finished before or during the general hackathon in a month's time."

(n2k8 hackathon summary to be continued)


Thank you to Mark for such a great account of the hackathon. We look forward to the rest of the series. Your donations and CD purchases fund these hackathons and make continued development possible.

(Comments are closed)


Comments
  1. By Anonymous Coward (85.130.30.230) on

    give us more more more :-D

  2. By Anonymous Coward (81.83.46.237) on

    What's up with the laptops? Sponsored by IBM? ;-p

    Comments
    1. By Anonymous Coward (2a01:348:108:155:20a:e4ff:fe2d:99ee) on

      > What's up with the laptops? Sponsored by IBM? ;-p

      Most things work properly with the older X and T series (30, 40), it's always nice to have good reliable suspend on a laptop.. Build quality was, on the whole, pretty reasonable too.

      Comments
      1. By Anonymous Coward (213.221.123.174) on

        > > What's up with the laptops? Sponsored by IBM? ;-p
        >
        > Most things work properly with the older X and T series (30, 40), it's always nice to have good reliable suspend on a laptop.. Build quality was, on the whole, pretty reasonable too.

        I totaly angree!

        But hopefully the situation changes for new models made by lenovo.
        They're still good but ACPI only (so no APM).

        Btw: Theo should have been placed into the middle of the picture (most importent guy is always the one in the middle) :]


        And I hope they wont kill any blowfish if they're in japan! Rescue the fugu 'course he ate all my svnds and passwords! :D

    2. By Fábio Olivé (201.14.141.44) on

      > What's up with the laptops? Sponsored by IBM? ;-p

      The build quality of IBM (hope Lenovo as well) laptops is awesome. They are really built to endure massive (ab)use by furious developers. The keyboard layout and feeling is great too. In summary, they are worth the price in every detail. :)

    3. By Mathieu Sauve-Frankel (222.0.65.5) msf@openbsd.org on

      > What's up with the laptops? Sponsored by IBM? ;-p

      It's a pretty well known (STFA) that OpenBSD developers have strong preferences for thinkpads, especially the X40 series, largely because up until recently IBM was the only vendor that had committed to continue supporting APM, which most us feel is still superior to ACPI.

      Now that lenovo have changed all of that, some of us have started hoarding
      older IBM laptops :-P, like T40 and X40 series.

      I got a chance to check out Bob's new X300 while he was here in Japan
      and I really didn't like the quality of it at all. IMO, Lenovo have done a great job of killng the thinkpad brand. Not sure what my next laptop is going to be.



      Comments
      1. By phessler (phessler) on first undead, then not, then undead again.

        > > What's up with the laptops? Sponsored by IBM? ;-p
        >
        > It's a pretty well known (STFA) that OpenBSD developers have strong preferences for thinkpads, especially the X40 series, largely because up until recently IBM was the only vendor that had committed to continue supporting APM, which most us feel is still superior to ACPI.
        >
        > Now that lenovo have changed all of that, some of us have started hoarding
        > older IBM laptops :-P, like T40 and X40 series.
        >
        > I got a chance to check out Bob's new X300 while he was here in Japan
        > and I really didn't like the quality of it at all. IMO, Lenovo have done a great job of killng the thinkpad brand. Not sure what my next laptop is going to be.
        >
        >
        >
        >

        I've had the t42, t43 (both ibm), t60 (lenovo), and a dell d630. I like the dell better than any of the thinkpads. also: built-in cereal :D.

  3. By giovanni (87.2.60.254) on

    naaa! I do not see their feet

  4. By Martin (130.225.33.69) on

    Happy Birthday, Theo. 40 years and still going strong.

  5. By Anonymous Coward (24.37.242.64) on

    Why doesn't the DHCPd syncing work over PFSync or such? As is now, does the dhcpd.conf manually have to be copied to the remote hosts(s), say via scp/ssh on each update or every X amount of time?

    On a side note, who's the lady on the left? I don't see mention of her, she's good looking. =)

    Comments
    1. By mk (193.128.72.68) on

      > Why doesn't the DHCPd syncing work over PFSync or such? As is now, does the dhcpd.conf manually have to be copied to the remote hosts(s), say via scp/ssh on each update or every X amount of time?

      pfsync runs in the kernel, dhcpd runs in userland. Besides, pfsync is for syncing pf state information, not a general synchronization interface.

      Comments
      1. By Anonymous Coward (74.14.136.206) on

        > > Why doesn't the DHCPd syncing work over PFSync or such? As is now, does the dhcpd.conf manually have to be copied to the remote hosts(s), say via scp/ssh on each update or every X amount of time?
        >
        > pfsync runs in the kernel, dhcpd runs in userland. Besides, pfsync is for syncing pf state information, not a general synchronization interface.
        >

        Time to make it syncd? Like how ho state d became relayd? I wonder how I will keep the state of my hos now?

        Comments
        1. By Anonymous Coward (24.37.242.64) on

          > > > Why doesn't the DHCPd syncing work over PFSync or such? As is now, does the dhcpd.conf manually have to be copied to the remote hosts(s), say via scp/ssh on each update or every X amount of time?
          > >
          > > pfsync runs in the kernel, dhcpd runs in userland. Besides, pfsync is for syncing pf state information, not a general synchronization interface.
          > >
          >
          > Time to make it syncd? Like how ho state d became relayd? I wonder how I will keep the state of my hos now?
          >

          That's exactly what I was thinking, then would come other ideas good things after this too.

        2. By mk (130.225.243.71) on

          > Time to make it syncd? Like how ho state d became relayd? I wonder how I will keep the state of my hos now?

          I suspect moving pfsync out into a userland process would be really expensive. Lots of context switches and app<->kernel copying.

          Right now we have a few userland applications (spamd, dhcpd, perhaps more) that synchronize state. Perhaps some of this could be extracted into a library or perhaps even a daemon but I doubt it's worth the effort.

          Note that synchronization isn't just synchronisation. If it were, dhcpd etc. would probably have used rsync.

    2. By Anonymous Coward (216.84.37.163) on

      > On a side note, who's the lady on the left? I don't see mention of her, she's good looking. =)

      Why would you do this? I see several good-looking guys there and don't feel the need to comment about it! I don't think she is a developer, but still.. this behavior makes female developers uncomfortable and less likely to show up at such events. Don't do it! ;-)

    3. By Anonymous Coward (79.114.19.131) on

      > Why doesn't the DHCPd syncing work over PFSync or such? As is now, does the dhcpd.conf manually have to be copied to the remote hosts(s), say via scp/ssh on each update or every X amount of time?
      >
      > On a side note, who's the lady on the left? I don't see mention of her, she's good looking. =)
      >

      http://zeus.theos.com/deraadt/t/c2.jpg

      Comments
      1. By Matthew Dempsky (76.235.124.214) on

        > http://zeus.theos.com/deraadt/t/c2.jpg

        Obviously photoshopped. :-)

  6. By Anonymous Coward (74.59.190.33) on

    Now I just wish you could use dhclient on carp interfaces...

    Comments
    1. By Anonymous Coward (24.37.242.64) on

      > Now I just wish you could use dhclient on carp interfaces...

      I agree, would be nice. There are other ways around it that do nearly the same though.

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]