Developers’ Weblog

Sponsored by
HostEurope Logo

Developers’ Weblog

All 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

FOSDEM Tag 1

07.02.2009 by tg@
Tags: event

Das Aufbauen verpaßt, aber wir haben ein „m“ (Bild wird später nachgereicht, sollte smultron freuen), viele Kontakte, und die englischen Flyer sind jetzt schon alle, die CDs runter auf ¼ oder so.

Der Unicode-syscons-Vortrag war für mich leider nicht so ertragreich; dadurch, daß wir vt100 wscons(4) haben, und durch meinen script(1) -lns Hack, haben wir schon mehr Probleme gelöst und Wissen angebaut als er. Ed Schouten ist aber anscheinend ein vielversprechender talentierter Jungentwickler.

Cool, ich habe ein bißchen WLAN! Mal schnell ein bißchen wlog Einträge verfassen, Benny und gecko2 wollen ja nicht.

Jetzt nur noch den NetBSD®-Kollegen neben uns zum Installieren des RANDEX-Plugins verlassen…

Wer setzt uns eigentlich direkt neben OpenBSD? Zum Glück gibts eine große Barriere, daher ist bislang, außer Laserpointerattacken (sogar direkt in Bennys Auge) noch nichts passiert…

FOSDEM Tag 0

07.02.2009 by tg@
Tags: event

Gestern, Freitag Abend, war der Tag 0 der FOSDEM. Natürlich waren wir – Fabian Köster und der Vortragende Robert Schuster, gecko2@ und ich – beim Beer Event, später auch mit bsiegert@. Das Bier war lecker, allerdings habe ich zwar nicht zu viel aber wohl zu varietätenreich getrunken, sodaß es mir in der Nacht nicht so wirklich gut ging und ich noch Bauchschmerzen habe. Also keinen mit dem Debian Projektleiter trinken.

Benny hat lustige Sympathiën von Leuten aus anderen Projekten bekommen, aber ich darf leider nicht drüber schreiben.

I’ve started trying to use the mircvs://contrib/code/heartbeat/ stuff to monitor NTP timedeltas between my boxen and a reference box (some random Stratum 2 pool server I do not use as server in any of the boxen, otherwise I might have used the PTB servers). Add rrdtool and rrdgraph output. Maybe mail when the boxen are down, until we have company monitoring set up?

I wonder if I should do it the “right” way instead of the “little effort” way, then commit it? Including cleaning up the age-old code. Is there any interest?

On a side note, we need a monitoring and management system, either one tool or integrating a few. It should have a command line interface and a WUI, different web pages for admins and (read-only) users who can look there for the general system state before complaining. Also, we need configuration management. A few keywords: nagios, cacti, puppet, cfengine. These were thrown into the room. Does anyone have a complete solution, possibly with VM management (how much does OpenCRM do?), for which we currently use a homegrown Jabber bot (don’t ask…) which does the template cloning (zfs, iSCSI) and other setup. Other suggestions, tools to avoid, success stories, links, documentation welcome.

Leute, kauft euer Nahverkehrsfahrkartenabonnement bei der Deutschen Bahn. Nicht nur nerven sie täglich weniger, sind ihre Züge in der Regel pünktlicher, sondern sie überfahren einen auch nicht fast… wie bei gewissen Stadtwerken vorkommt. Und in der Regel halten sie sogar da, wo sie sollen.

Auf jeden Fall habe ich meine innerstädtische Monatskarte bei der DB gekauft und hatte heute einen Gutschein für ein Jahr BahnCard 25 im Briefkasten. Nett…

kdm starts while (Ctrl-Alt-F8) the system is still booting. I only noticed it because we log in against LDAP (UCS), and since I have two NICs in there now (due to FreeWRT stuff), the Network Manager takes more time in bringing the network up. Gah. I do not even want to log in before /etc/rc.local has run, because it contains calls to rdate and starts an egd (entropy gathering dæmon), among other things… these cheaters are worse than Microsoft®!

Oh, and when I wanted to unlock it to write about it, it hung (as if it didn’t like me). Alt-SysRq to the rescue (we need it in MirBSD too). Crapware.

MirGRML and FOSDEM DuaLive™ Triforce™ CDs

03.02.2009 by tg@
Tags: bug event grml security snapshot

Uh-oh, 03.02. already. I think, after the switch to GRUB 2 and another couple of bootlooter fixes, that we’re there yet. Funnily I only noticed how BSD cannot access labelled disks when the device’s sector size mismatches the one in the label. There is also an embarassing (for upstream) local DoS exploit possibility, by setting a sector size of 0, the kernel traps division by zero. Thanks for all the blowfish, yeah.

The checksum file for MirGRML 2009.01 (experimental, but probably, by now, complete). You’ll probably figure out the ISO link.

The full MirOS CD should be done soonish as well. Just have to test it, then I can go to bed. Wish me luck.

GNU grub-legacy cannot access ISO 9660 on devices with sector sizes other than 2048 bytes… and, according to mika, has other quirks, with recent mke2fs’ inode defaults making it hiccup. Heh.

more bugs – bootloader and mksh

01.02.2009 by tg@
Tags: bug event grml mksh snapshot

Again. I almost have a MirGRML+bsd4grml ISO ready (exactly 72 MiB), just for the fun of it, and so that people can toy with it – and test the integration. But nooo, I even hacked a disklabel, yet it wouldn’t access the filesystem. The “machine label” command shows what’s wrong (and hints how to fix it), and I missed a corner case in disc I/O due to adding two not-so-independent-from-each-other scenarios during the El Torito merge.

mksh has funny behaviour with ${foo/@(%)/x} failing – only in Unicode mode though.

ext2fs damage

01.02.2009 by tg@
Tags: bug

Fun: MirBSD #8, #10 and #10-current can’t edit or create symlink(7)s on ext2fs, MirBSD #7quater could. They hang the filesystem, as well as the parent of the mount point, on access, and damage the filesystem. I have no idea… (msdos is damaging long filenames, btw.)

4.4OpenBSD can… oO

Benny has done more work on ext2fs, maybe he’ll fix it.

more bootloaders *grml*

31.01.2009 by tg@
Tags: bug event mksh snapshot unicode

To counter the bug in Parallels Desktop, I rewrote all of our Master Boot Record (mbrldr, mbrmgr), Partition Boot Record (bootxx), and BIOS disc detection code in boot(8/i386). I've also changed the magics, API etc. a little between these, rewriting or removing quite some parts of both installboot(8/i386) and bootxx.sh a.k.a. self-installing bootblocks. Oh, and MBR and PBR are smaller, or rather, have more room for informatory texts and data block storage now.

Lucas “laffer1” Holt from that cat's BSD helped me testing, qemu and bare metal IBM X40 works, so I suppose it's good. Can't test on VMware Server 2 right now.

On the other front, I'm writing this wlog entry on MirGRML, which is finally sort-of finished (which is why it has much less Unicode than a regular posting from my laptop). lynx-cur in sid is broken though, due to it using GnuTLS. Gah! Anyway, I see a sid “wtf” repo coming, and it was workarounded; next time I'll just do things differently.

Lukas “smultron” , the graphician of our partner project mnbsd, helped a little with the label (while I'm not versed in that field, if I had gotten the right material to work with I could've managed it all alone this time), and all that's left now is the Live-CD part and baking (and testing! even on sparc...) the ISO 9660 image.

There will be 297 MirBSD Triforce™ DuaLive™-CDs (two will go to Kiwi land to swishy, one is already reserved for XTaran) and 50 MidnightBSD Live DVDs at FOSDEM.

gecko2@ and bsiegert@ were not too helpful today, although I have to admit that real life often has precedence. I merely read some and went to sleep early yesterday too (and disabled both POTS and cell phone so that I couldn't be waked, haha!) instead of hacking late. Luckily, the deadline is not that dead, and I'll almost certainly make it tomorrow, in time. Thanks again, Daniel Seuffert from AllBSD.de, for all support you and others give to the smaller BSDs. Heh, and laffer1 is not quite done yet with his ISO (or UDF?) either.

2009-01-24 snapshot uploaded, pending validation

25.01.2009 by tg@
Tags: grml snapshot

Yesternight, I built an i386 snapshot (including a bsd4grml). This is already uploaded on /MirOS/current/i386/ but beware, it is pending validation, and some things (like the serial console or combined i386+sparc ISOs) are missing. Note that there will be, due to time constraints, no new sparc snapshot, we’ll just use the last one I built.

If people who actually use MirOS/sparc were to talk to us, this situation might improve.

It’s still a couple of days until the Triforce™ ISO and CD label are due, but we’d appreciate independent test results, as usual.

Plans for the FOSDEM Edition Triforce™ (MirOS DuaLive™ + MirGRML) CD-ROM are to add a few packages to the usual baseLive image: IceWM, screen, rsync(GPLv2), mc; possibly ent, lzmadec

Watch the news for announcements about usability of the 2009-01-24 snapshot.

Update: the three serial console floppies and the five ISOs are there and should arrive at the public mirrors tonight. They are in fact bootable and quite probably usable, but bsiegert@ still has to do the validation by doing a full install and some port builds.

Das Wetter bessert sich

25.01.2009 by tg@
Tags: geocache

Heute mal wieder das Vélo aus- und mich gut eingepackt:
OpenCaching.de The Frog Site Rare Jewels Broken Webpages²
        inc     dword ptr ds:[caches]

Diesen hatte ich auch schon länger auf’m Kieker, allerdings nie die Muße, dort hinzufahren, bei dem Wetter, und der FTF lockte auch nicht, weil er – mal wieder – auf der Kommerzseite Tage bevor er auf der freien Seite zu finden war geloggt wurde. *grml*…

Liebe Plänet Sümlink Leser, weil so viele von euch zu <zensiert /> sind, sich das CAcert.org Root CA Certificate zu installieren, und die Bloatzilla-Leute ihren A^H^H^H^H Hinterteil auch nicht bewegt kriegen, editiert der XTaran das https in ein http. Zu dumm, daß ich auf vielen Seiten nur noch SSL erlaube. Nutzt also den „Mehr...“-Hyperlink, um die Statistikbildchen zu sehen, und hofft, daß die neue CAcert-Root bald live und in die Brauser geht. MirPorts Framework Nutzer habens natürlich einfacher…

Gleich geht’s erst mal zu meinem Stammrestaurant, einem Jugoslawen, ääh Kroaten (Primošten oder auch Dalmatien Grill) mit bogus…

miscellany

24.01.2009 by tg@

My pcc wchar_t patch thread gets attention again. Yay!

Good I just installed MirOS on soon-to-be-tear (althouh on odem’s old 2.5″ HDD. I cou^H^Haught an error due to /dev/MAKEDEV using the installer’s user and group databases, which were missing some. And good I compiled a kernel beforehand… bzero(3) is optimised in locore, memset(3) isn’t. Lession learned? Maybe. I bet I’ll do quite some of these “Flüchtigkeitsfehler” again in the future, I always used to do so, even at school.

Damn. SCSI is expensive, and SSDs are available only for SATAn.

Whew, finally weekend. I slept what appears to be 10 hours and dazed for some more. Now I’m unwilling to do any work, but I’ll do what I’m owing y’all. (Yes, normally I try to write British English.) I estimate we’ll have a new i386 snapshot, and mika his special edition, by tomorrow evening. Too bad I delayed testing the MirGRML due to the dayjob workload and general… unwillingness. Anyway, FOSDEM is coming, we’ll shine.

Lucas “laffer1” Holt from our other ally MidnightBSD will also do a special ISO for FOSDEM, which we’ll distribute among people. I wonder if Debian GNU/kFreeBSD does some, maybe an updated GING CD, but I guess that’s up to the Debian people, not for us BSDers.

Since I started working, I can’t bring myself to eat anything in the morning, nor really in the evening. I wonder why that is… I still have a package of black bread here I bought when gecko2@ visited me, that would be 31 Dec I guess. I probably have to throw it away by now, but that sucks. I hope I don’t get even more ill… my colleagues and I have not been feeling all that well for days.

On building a cross-compiler package: I wish I could bring myself to get it done by FOSDEM, maybe Robert Schuster would help baking the JDK.

Anyway, I’d better stop concocting some text here, and instead bring the mock-tear installation into working state and get it to start rsynching, cvs(1) co(1)ing, and compiling…

All 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

MirOS Logo