Slashdot Mirror
OpenBSD Hackathon Underway
Triumph The Insult C writes "Aside from some stealth developers, the annual OpenBSD Hackathon, held in Calgary, is underway, according to Theo. They've been doing some recent work on SMP, and have some impressive AMD SMP gear there that they've got to hack around with. A few years ago, it was PF. Who knows what they'll come up with this time that knocks our socks off."
If history repeats itself, it'd be an Apache clone.
These hackathons always scare me a bit. Major functionality has a habbit of going from non-existant to solid before it's over.
I rarely criticize things I don't care about.
Accordingly, word on the street is that significant effort this hackathon will be put into fixing the first ever OpenBSD virus...
I think a fix has already been found for this particular "virus".
Want to improve your Karma? Instead of "Post Anonymously", try the "Post Humously" option.
It's at times like this that we need a "-1: Idiocy of post only matched by idiocy of moderation" option for moderation.
# fstat -f / | ipsecadm - - | tcpdump -i - | less
Thanks for your contribution to my state of the art, nacturnation (do you mind if I call nacty?)
--
With all the work that has been done on SMP, hopefully they'll get some good work done on it. miklas@ has done some wonderful work, while I may not agree with the big kernel lock, it's a start.
/. instead of hacking SMP up there :-p
Good luck to those reading
Error 407 - No creative sig found
The record is clear on one thing: no operating system has ever come back from the grave.
Oh yeah? How about Amiga? How about BeOS? Hm... Never mind.
I installed 3.5 yesterday, oddly enough.
It only took 15 minutes to install the base system. All my hardware autodetected without problems. What's wrong with the installer?
(hint: it's humor, laugh)
i cant emagine an easier installer. i have yet to see a single os that can be installed as fast as openbsd without fast clicking the next and default buttons on some gui POS.
its just that most people dont understand how a hard disk works, and if you cant take the time to learn fdisk and disklabel, you probably wont be able to take the time to learn how to use an operating system without newbie user abstraction. openbsd is free, functional, and secure. i dont think ive ever seen any reference to advancement in mickey mouse hand holding techniques.
put two new users that understand OS concepts in front of a FreeBSD installer and an OpenBSD installer and see who goes from zero to puffy from a floppy before the other guy can start his package extraction.
I did a network install on an old laptop, and once it got going, it was fine. However, the first time I tried to install it froze in downloading some packages, and there doesn't appear to be any way to cancel out of that, or anything to detect that there's a problem and back out. Not a huge deal - I just started over.
Better resumability/recoverability during the install would be nice, and the ability to easily build custom bsd.rd's that run your own install script/whatever for mass provisioning, would be nice.
I would like semi-automatic security updating like you can get with Debian's apt-get.
demi
Fishing season is over so you can stop the trolling already and get back to your comfy win98 box.
I caught you didn't I? Moron.
I actually hope they don't make the installer more user friendly: otherwise we'll have too many supercifical and clueless users coming to use the system. OpenBSD (and NetBSD is a bit like this as well, more in contrast to FreeBSD and Linux) tends to be directed to knowledgable technical users, which goes in tandem with its security ethos. If you like nice installers, try another O/S: OpenBSD isn't reall for you.
I'd much prefer them to be using their time on innovative security features, not pretty installers, SMP, apache-clones, etc. CARP, pf, privsep, etc.
In fact, the OpenBSD guys would probably like it people with good ideas raised them here, and who knows, these ideas may make it into current or future release.
Actually all users could use a thoroughly code-reviewed system with a safe default install.
The net would be a bit safer for sure, even if it was just a bit.
More user-friendly installer => wider user-base => less zombies for DDoS. Maybe even more money for OpenBSD development? More OpenBSD related jobs? More interest in embedded ports? More positive PR?
Too bad you're so shortsighted!
release(8)
site.tgz,[install|upgrade].site (faq 4.13)
vodka, straight up, thank you!
Thanks, but it doesn't really address any of what I asked for. The man page you reference describes how to build a release--that isn't what I need since I can mirror the releases just fine. site.tgz doesn't help, because you still have to have someone sitting there partitioning the disk, selecting the install URL, selecting the disk sets, plus the site.tgz set, and doing other monkey-activities that you can avoid with a good kickstart disk--these are what I would want to automate through a custom bsd.rd. I understand that this is probably already possible--I'm asking for it to be convenient.
And if you're meaning to imply that I should apply a source patch, rebuild a release and then go through the reboot-upgrade cycle on 400 machines, in order to apply a security update; well, that would be impractical. With Debian, on the other hand, this couldn't be easier.
demi
I a few weeks ago, I decided to give OpenBSD a try, having used FreeBSD and Linux quite a lot. So I downloaded and wrote the ftp install disk and booted it on my old AMD Athlon 750. I set up all the disks slices and everything was going fine until it was time to initialize the NIC so I could start downloading packages. The install issued an error message about a "stray interrupt" on IRQ 7 and then it locked up hard. So, I thought "that was strange" and power cycled the machine and tried again. Same thing. So I googled through countless message boards and determined that the kernel probably didn't like the chipset on my motherboard. There were various fixes offered (such as enabling lpt in the bios) but none worked.
Now the interesting thing is that a vanilla linux kernel (I built Linux From Scratch on this box before) will issue a similar message ("spurious interrupt"), but Linux won't lock up like OpenBSD did. Linux just sends the message to stderr and keeps on working. Most distribution tweaked kernels don't have any problem at all and never complain.
Well, "no problem, you've got some flaky hardware" you are probably getting ready to say. I thought so too. So I tried it on another box (This one a P-3 450 Compaq). This time the installer wouldn't even boot. It issued some cryptic message and stopped before the installer's kernel was even booted. Having struck out on two machines, I gave up and installed FreeBSD without incident.
I had started with high hopes of giving OpenBSD a whirl because I wanted to try out what has been described as a bloat-less, minimal, well put togather unix that emphasized correctness over bleed-edge features. I was not impresed. The whold affair has sort of put me off from OpenBSD. I know OpenBSD is focused on security and not having pretty installers. I don't care if the installer is pretty, but I do care it it locks up and is unusable on fairly standard equipment. Neither of these machines have anything exotic about them. They are just standard x86 boxen that work well with FreeBSD and Linux. The OpenBSD team needs to give their installer some more polish (from the functional standpoint not the prettiness standpoint)
Christ. I know this post is offtopic also, but what is the deal and all of these people bashing FreeBSD. I mean... if it's actually dead, new updates wouldn't be coming out, would they? You do know that FreeBSD 4.10 was released recently, don't you? I'm going to be throwing together a computer this summer just for installing FreeBSD on. As long as software has some users and continues to be updated, it's not dead. Hell, Windows 95 isn't even dead yet... it may not be updated any more, but a lot of people still use it.
"Instant gratification takes too long." - Carrie Fisher
I believe i read that one of the points of this was going to be dramatically improving the fail-over abilities of OpenBSD.
"The crows seemed to be calling his name, thought Caw."
Actually usability and security often enough go well enough hand in hand.
The easier the default install, the less likely are errors. The more powerful the configuration interface, the less errors, the more safe installs.
Features like W^X or stack canaries make it harder to crack a box, but not impossible. But an app free of overflows and with a good config does. So add nice management/installation tools.
In fact, with the choice of already available installer apps out there it shouldn't be too hard adapting one, so what's the big deal?
Most of the security of OpenBSD stems from careful code review, not featuritis. But they won't be reviewing, I bet, but adding features.
Oh, and beside: I guess they'll mostly focus on SMP anyway. That's not a security feature at all, so your point is moot. Most of the current OpenBSD user population doesn't even have use for it, so it clearly is a feature aimed towards more popularity.
The OpenBSD team needs to give their installer some more polish (from the functional standpoint not the prettiness standpoint)
It seems you have been quite unlucky.
I've been using OpenBSD since 2.5. The installer was a bit of a shock at first, but once it makes sense, it is wonderful (and it is sensible). I usually can install OpenBSD with X in under 5 minutes and I've only ever found one machine to not install for me (an IBM Thinkpad series 1300).
I've installed on tons of x86 machines, some MacPPC, a 68k Mac and a Sun Ultra 10. No problems. I cannot say the same for some Linux distros or the other BSD's.
I plan to have another crack at that Thinkpad, now that OpenBSD 3.5 has new boot code.
There are times when OpenBSD won't work, where some Linux distro does or vice versa. So try again. Don't write OpenBSD off because of a few bad initial attempts. I've been using Linux for about 7 years and tried lots of distros and the BSD's. I settled on OpenBSD (and Debian when I must use Linux).
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
That said, I'm going back to Linux. Why? The scheduler: I really miss the ability to do a bunch of things at once and still keep playing Ogg files.
I've got a slower machine -- used to be an overclocked (450MHz! woo-hoo!) Celeron, now down to stock 300MHz after a bunch of crashes recently. I run IceWM, bunch of xterms, XMMS and Phoenix/Firebird. Under Debian, 2.6 kernel, I could do all of that without any problems. Same machine under OpenBSD will chug and stutter on Ogg files when I hit reload in Firebird, switch windows, or start a new xterm.
It's amazing how much a difference there is in the feel of the desktop. I'd choose OpenBSD in a heartbeat for a server, and maybe for a desktop in a faster machine, but I really like the feel of the recent Linux kernels; it makes all the difference.
Carousel is a lie!
rdsetroot to extract the fs image
vnconfig the fs image
mount it, edit the install scripts
umount, vnconfig -u and rdsetroot to stuff the filesystem back in.
takes 2 minutes to do and works extremely well.
what's a stealth developer? Do they dress in black and only program on black laptops in the shadows, where no one can see them?
Now you face my Ninja C++ operator of death!
Is some more info on what SMP(Symmetric multiprocessing)actually is and what it's advantages and drawbacks are.
Creative Demolition
I'm a die hard OpenBSD user since 2.5, but I have to agree that the scheduler (or whatever) is definately suboptimal for a desktop -- even renicing things as root doesn't help very much. Disk scheduling is much the same. Clearly it just hasn't been designed with desktop responsiveness in mind. This issue, plus software availability (things like OpenOffice) still keep Linux well ahead as far a desktop goes.
It is interesting that OpenBSD puts so much effort into painless soundcard support, and in the past X11, considering they otherwise seem to make few desktop concessions. Perhaps someday they will improve on the (apparently) server-esque scheduling, who knows?
I was using a 3.4 install disk. The Compaq that I tried it on wouldn't even boot at all with 3.4. If the boot code is different on 3.5, I'll give it another try.
Oh cool, I hope it works for you.
Sometimes when I create OpenBSD -stable UltraSPARC CD's, I boot from older genuine OpenBSD UltraSPARC media and then swap with my -stable CD (because I have put little effort into figuring out how to make a bootable UltraSPARC OpenBSD CD). One day, I rekon it might bite me. ; )
My Thinkpad is a type 1161-41U. OpenBSD 3.5 bootable CD (home made from -stable), I can confirm, still does not boot.
I might try booting from various 3.5 floppies...
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
That said, I'm going back to Linux. Why? The scheduler: I really miss the ability to do a bunch of things at once and still keep playing Ogg files.
This is surprising. I ran OpenBSD on a PII-300. I would run MP3's without any skipping, while surfing the net and making -stable release in a minimized xterm.
I have not tried the Linux 2.6 kernel yet, however, so I can't comment on feel.
Linux 2.4 certainly did skip MP3's on this machine.
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
I guess the only chance OpenBSD will ever have in the corporate world is on servers and firewalls, and in security-related embedded devices.
It's fine for this because it is designed for this and it clearly is the focus of its designers.
I cannot imagine a desktop OpenBSD simply because nobody is there who's willing to apply the necessary polish to make it worthwhile and usable in a user sense.
If no one is willing to contribute a better installer, while every Linux distro can offer one (even Debian is hacking on a better one), it's fairly sure no one wants one, or the people that could make one are deterred by some aspects of the OpenBSD project - perhaps on a personal basis, or simply because they think their work would not be appreciated, or because they think OpenBSD's tiny marketshare would simply grant them not enough fame or whatever rewards open source work.
I'm aware that the *BSD projects already are fairly well behind on attracting developers. I guess the Debian project alone has more developers than all *BSDs together. I guess there wouldn't be a ports tree else. If there would be enough maintainers, all would be packages.
So I'm not very surprised there are not enough people for writing installers or polishing the dektop. Yet a usable installer would greatly help OpenBSD popularity, leading to a greater developer exposure.