FOSDEM review and my extended outtime from Gentoo

I don’t know, where I should start, but FOSDEM 2007 was the best FOSDEM I have been to. From Gentoo’s view we had the most developers ever crowded at one event. Too bad that I again missed some, but it was nice to meet some of our new German devs, like phreak and jokey, and meet genone I ever missed before (at least at FOSDEM 2005). Then it was a big pleasure to meet the US-devs, starting with wolf31o2 and vapier, and again kingtaco.

Beside Gentoo I had a lot of talks with persons from the Free Software Foundation Europe as I know some of them personally for a couple of years already. Then I met some people from the CCC and had a rather long talk with Tim Pritlove about the future of Pentabarf, a very good tool for the organisation of a conference (and which I know from its early hours). And somehow I got into a situation where three different persons greeted me at the same time, but those three don’t known each other… Am I so famous? 🙄

Now let me collect things which might need to be improved in the future:

  • The booth (well, our tables) looked good at Saturday when we had some nice hardware to show, but on Sunday it was mostly understaffed and we had nothing to show. Furthermore many people asked for T-Shirts and give-aways. Time for an “event-kit”!
  • Too bad that there isn’t an area where you can sit together, have network and squash bugs together with another dev. I think that this is one of the things we can establish in the “Gentoo Village” at the Chaos Communication Camp this summer. Sitting in couches, having a beer and listening to ambient music while working together with users on their pet-bugs.
  • Dinner in a pre-noble restaurant which costs half of the money you spent for the whole weekend should not be repeated. It’s better to have a buffet where you have to go to now and then and have a talk with different persons while you fetch another drink, instead of talking to the same persons at your table for the whole evening as you are somewhat bound to your seat.

Now let me thank some persons, starting with Wolfram, who picked me up in Cologne and drove me back. Then Dimitry and Guy for organising the dev-room and booking the Hostel. It was so relaxing for me that I had not to do this once again like in 2005. And all devs who attended FOSDEM and made this an unforgetable event!

Currently I’m preparing my extended outtime. I finally have to finish my diploma-thesis, which will take about three to four months. During this time I will not be available for Gentoo work. There are still some bugs open on my buglist but I don’t think I can resolve them soon. I already quit IRC as this is a major time-burner. But as faster as I work on my thesis the sooner I will be back 😉 See you again this summer!

I do it one more time: Gentoo CVS move

The title sounds like I do it monthly… No! But it’s the second time for me that I move Gentoo’s CVS to a new server. I already did it in September 2003 when we moved from our former CVS-box (funny, I even can’t remember its name) to our current CVS-server, named “lark”. At that time it was a pretty easy task, as we didn’t had so many developers, a quite small CVS-tree and not so many scripts for automatic actions.

Nowadays lark provides beside the CVS for the portage-tree a couple of repositories for other subprojects and SubVersion for some more subprojects. We also have some scripts for backup, hooks for mailing the CIA-statistics or the changes of the English documentation for the translators.

Setting up a new server is a good time to review the scripts and make some things a little bit different. It’s good that we have the new server, called “stork” for two months already. So I had enough time to set up the backup differently than before. As we use LVM2 for our cvsroot- and svnroot-storage, I can quickly create an lvm-snapshot of the current tree and create a backup from that snapshot. The old scripts did a live-backup and sometimes it happened that somebody commited something during that time, so that the backup was not that clean… The creation of the backup takes about 30 minutes (mostly it’s bzip2 which takes so long). And transfering the backup to our backup-server takes about the same time.

robbat2 worked a lot on the ssh-key-transfer. Before we copied the authorized_keys from our dev-box to lark on a hourly basis. Now we use LDAP (finally) to extract the ssh-pubkey, which takes only two seconds (woot!). So we switched to a sync every 15 minutes, which will make devs more happy!

At 7UTC on Friday we will take down lark and do an rsync of the tree to stork. A test a couple of days ago was done within three minutes. Both servers will be on the same state then. Let’s hope, that the DNS-update will also be fast. We have an estimated timeframe of two hours for the move, but I guess, we will be ready sooner.

Some words about the machines:

lark

  • active since September 2003 (that makes 3