FreeBSD to rethink target audience?

Mel_Flynn! Yes, I have tried PC-BSD and it's a great desktop-setup of FreeBSD. Every thing works out of the box! But it uses KDE and I don't like KDE. :e And I think I'm not the only one. So I try setup my own FreeBSD with Gnome. Not easy but I have no choice. ;)
 
If you setup your desktop the same way every time, then write a script. I have a post-install script that installs on my programs on FreeBSD and restores my configuration files. It runs beautifully.
 
chrcol said:
The 2 top complaints I heard from datacentres as to why they dont/stopped supporting FreeBSD.

1 - Hard to use installer.
2 - Fussy on hardware, hangs on bootloader are common.

My view on hardware compatability and drivers, well for a long time on realtek cards linux was better, the reason I got from FreeBSD when I queuried was its poor hardware so tough luck, rather than ok we will try and get our driver as good as linux driver. (...)

Hmmm. I have had my issues with Linux, and much of it related to installation. First, with Linux, none of the distros I tried (SuSE, Fedora etc) documented what hardware was actually supported -- the only way you would find out is by installing and then noticing that Linux didn't recognize your RAID card. Yes, there were other drivers out there, but the Linux installers didn't give you the option of inserting a floppy or CD with additional drivers. A day of messing about with this stuff, and I was already fed up with Linux.

Later ... the redundancy of the RAID array was foiled by the much-vaunted EXT3 filesystem corrupting its journals twice. (I since downgraded to EXT2, and it seems to have been OK.)

This server will shortly return home, with a clean install of FreeBSD 7.0 (I was hoping for 7.1, but it seems not imminent), and I will most definitely not miss Linux. Otherwise all our servers run FreeBSD (though some of them really do need to migrate from 4.x to something more current).

But 100 million Elvis fans can't be wrong, can they? :\
 
BSD != Windows

Mel_Flynn said:
No, it's a security disaster. Uptime actually means you didn't patch, it doesn't have any real relation to stability.

If we were talking about Windows (which needs a reboot with almost every security update) or Linux (which has lots of kernel updates requiring reboots because they cramp everything into the kernel), I'd agree. But actually I cannot remember any vulnerability in the FreeBSD kernel that's affected us in the last five years or so. Edit: Of course, our machines all run on customized (read: stripped-down) kernels, so YMMV.

Some packages such as Apache, MySQL, OpenSSH, Ruby, Rails and so on need continuous updating, but that doesn't require a reboot, only a SIGHUP.

So I don't see the long uptime for FreeBSD boxen as a statement on their security. They're not Windows or Linux after all.

Out of curiosity, for which security update did you last reboot your FreeBSD boxen?
 
Weinter said:
That is were you are wrong

Market share give controlling power so companies have to consider FreeBSD as a platform to roll out their ports

...

I can't agree with u more
 
thortos said:
If we were talking about Windows (which needs a reboot with almost every security update) or Linux (which has lots of kernel updates requiring reboots because they cramp everything into the kernel), I'd agree. But actually I cannot remember any vulnerability in the FreeBSD kernel that's affected us in the last five years or so. Edit: Of course, our machines all run on customized (read: stripped-down) kernels, so YMMV.

http://security.freebsd.org/advisories/FreeBSD-SA-08:03.sendfile.asc
http://security.freebsd.org/advisories/FreeBSD-SA-08:04.ipsec.asc
http://security.freebsd.org/advisories/FreeBSD-SA-08:07.amd64.asc
http://security.freebsd.org/advisories/FreeBSD-SA-08:08.nmount.asc
http://security.freebsd.org/advisories/FreeBSD-SA-07:01.jail.asc

I could go on, but I think you get the picture.
 
Mel_Flynn said:

I do, but I also see that there aren't many systems which are affected by any of those vulnerabilites, and most of them require local access of some kind. (If somebody has local access, you're fscked anyway. And what is that about a vulnerability that only affects systems with writable-only files *and* the attacker having local access? There must be more people with three legs than machines vulnerable to this.)

If you compare the scope of those things with the scope of the average Windows vulnerability, you'll notice a difference. The same is even more true for the total number of vulnerabilities, just look at the archives of the FreeBSD security mailing list and count the incidents there.

I'll admit that the IPSEC DoS is nasty, but that only affects VPN gateways and the three people using IPv6.

I'm preaching to the choir, but to sum it up, in the years I've been using FreeBSD (since the 4.x days), security of the base system has been a non-issue. It helps that FreeBSD both is very secure already in the standard configuration, and is pretty easy to lock down even further.

In that light, your statement that long uptimes say something about the security of FreeBSD systems is really wrong. If you have sufficiently locked-down boxes, rebooting for updates simply isn't needed.

You haven't answered my question, though. For which vulnerability have *you* had to reboot a system, and how often have reboots occurred at all for you?
 
That is were you are wrong

Market share give controlling power so companies have to consider FreeBSD as a platform to roll out their ports
...
I can't agree with u more

Some of them disagree with me...

I do admit dumbing FreeBSD down may be detrimental to the quality
But i feel support for latest hardware should always be a priority
To increase marketshare new system will always have new hardware
Maybe the dumbing it down maybe PC-BSD's job
But ultimately hardware support will still fall flat on FreeBSD shoulders as PC-BSD is not doing anything about it...
 
I kicked my addiction to uptime long ago. I don't even attend AA meetings anymore.
I reboot whenever I feel it's necessary. This can range from kernel patches to "making sure the boot sequence is still working correctly after adding a new service to the mix". I also went to single user mode, after the libc update earlier this year.
My approach to the problem is just different then yours. You assess kernel patches based on the workload and software of the machine. If I'd do that, I would feel compelled to run a syscall check each update of all software, to make sure they don't use the known broken kernel interface in the way that it's broken. This isn't very economical to me.
So, I make sure my syscalls (and userland) are "working as they should, to the best of my knowledge". Then I only have to deal with the software, within the context of a jail as I've limited the risk of kernel panics or hangs to a minimum.
As you've probably figured out now, the last security induced reboot was the nd6 patch, even though I don't route ip6 anywhere (yet).
 
Weinter said:
...

Maybe the dumbing it down maybe PC-BSD's job
But ultimately hardware support will still fall flat on FreeBSD shoulders as PC-BSD is not doing anything about it...

Well, as you said -that's not PC-BSD's job.

PC-BSD is FreeBSD with a fresh coat of paint. You wouldn't ask a house painter to fix your roof or work on your pipes.
 
snes-addict said:
There is nothing wrong with the FreeBSD Project focusing on servers. Looking at the recent improvements that the team has made, we see lots of talk about SMP and the new scheduler. Performance improvements such as these are absolutely vital for FreeBSD to remain a top contender for the best server OS, and with many desktop users on multi-processor machines, they also happen to benefit the desktop as well.

Also, one of the aspects of FreeBSD which I have always liked has been the cleanliness of the system on an initial install: the included software is simple and doesn't bloat. Seeing as how desktop users typically have everything from word processors to web browsers to video games, the addition of extra software to meet those requirements would make the FreeBSD cumbersome and particularly unusable for the server and cluster guys.

FreeBSD is already the best OS; perfect for servers, and (with the help of ports) is a great choice for the desktop.

indeed, FreeBSD made great strides on threading in 7.0 for mysql server on FreeBSD, that change alone suddenly made it a lot more viable for web server use, then add the new scheduler, and the removal of some thread safe code (due to buggy SMP in 5.x and 6.x) and suddenly FreeBSD is looking strong again.
 
tomh009 said:
Hmmm. I have had my issues with Linux, and much of it related to installation. First, with Linux, none of the distros I tried (SuSE, Fedora etc) documented what hardware was actually supported -- the only way you would find out is by installing and then noticing that Linux didn't recognize your RAID card. Yes, there were other drivers out there, but the Linux installers didn't give you the option of inserting a floppy or CD with additional drivers. A day of messing about with this stuff, and I was already fed up with Linux.

Later ... the redundancy of the RAID array was foiled by the much-vaunted EXT3 filesystem corrupting its journals twice. (I since downgraded to EXT2, and it seems to have been OK.)

This server will shortly return home, with a clean install of FreeBSD 7.0 (I was hoping for 7.1, but it seems not imminent), and I will most definitely not miss Linux. Otherwise all our servers run FreeBSD (though some of them really do need to migrate from 4.x to something more current).

But 100 million Elvis fans can't be wrong, can they? :\

I also use FreeBSD on all my servers :) even when I have to pay for a remote KVM to install it myself.
 
Mel_Flynn said:
Barnie said:
I think that FreeBSD has all components, tools, drivers and quality for desktop use! It's only the last step that is missing... setup/configuration for desktops.
This is why putting the effort in PC-BSD is a good idea.
It is a good idea in some ways, but it may give people the impression that KDE is a sort of default desktop for FreeBSD. Some of you are apparently comfortable with that idea, but many of us are not.
Mel_Flynn said:
This thread is about changing focus of FreeBSD itself and I don't think it's good for the project.
I don't see how providing a canned desktop setup would change the focus of the project. Practically all users of FreeBSD - even those who complain about the silliness of GUI - use at least X with one of the simple window managers. So let us make a canned desktop setup with one of those window managers. This approach might even attract the interest of many of the real experts, who for the most part use simple window managers themselves. Such a setup done right would also be a neat way to answer a number of questions that keep recurring on forums and mailing lists.
 
softblur said:
It is a good idea in some ways, but it may give people the impression that KDE is a sort of default desktop for FreeBSD. Some of you are apparently comfortable with that idea, but many of us are not.
No. It's the desktop that is easier to make and keep working correctly. Donate resources if you wish another desktop.

I don't see how providing a canned desktop setup would change the focus of the project.

I don't see that either. Especially since I never said that. Read the original post of the thread. That's what I commented about. Providing a solid desktop and mobile OS requires much more then a few scripts that install a desktop after OS installation.
 
sremick said:
PBIs and the choice of KDE is what keeps me off PC-BSD.
Hem... PBIs are not a must. You can furthermore use the original ports and packages. So this is not a real argument against PC-BSD.

Of course the KDE is a must... and this could be a contra argument.
 
Barnie said:
PBIs are not a must. You can furthermore use the original ports and packages.
I'm aware. But a lot of the value-add of PC-BSD are the GUI tools for PBI management. My point is, I would've preferred the same tools, interface, and infrastructure built up around a standard ports/packages type configuration, versus a wholly separate method that runs contrary against standard Unix-style software installation philosophy.
 
What is a canned desktop setup? I'm in favor of FreeBSD adding additional common sets to make desktop setup faster (I did not say easier for a reason).
 
ninjaslim said:
What is a canned desktop setup? I'm in favor of FreeBSD adding additional common sets to make desktop setup faster (I did not say easier for a reason).

In short, that if you choose the "X-*" distribution set from the installer, that you'd get a menu where you have to pick a desktop and that the display manager would be configured on ttyv8 on the next boot.

With a DVD install, this would be possible, from CD-ROM there isn't enough room on the primary disk.
The problem is where does it stop:
  • Should cam be configured so that GUI dvd writers work out of the box?
  • Should devfs.rules be modified with a designated group for desktop use, that gives group permissions to removable media?
  • Should nv driver be used or nvidia, if so detected? Same for radeon?
 
Uptime counts are just pecker contests anyways.

But in all reality, folks...would you really WANT FreeBSD to be a "mainstream" OS? Look at all the Linux communities across cyberspace, full of half-wits who jump on the Linux bandwagon because "M$ IZ TEH 3V1L!". The bigger your base grows, the more of these kinds of people you attract. It's one of the major reasons I can't stand to run Linux locally; looking for help on something requires me running into about an 8:1 noise:signal ratio on forums and websites.

I honestly do not like the prepackaged versions of BSD, namely PC-BSD and DesktopBSD. They sound like great ideas at first, but they seem to really limit the system. Sure you can make major changes to the system, but if you do that, you might as well just use the original system they're based on.

And that thing about techs saying FreeBSD is "too hard to install" is just plain laziness. Even my wife knows her way around sysinstall and can configure just about anything in FreeBSD, and she's in public relations.
 
I think PHK once said something like: "I don't care if FreeBSD has only 1% market share, as long as we have the best 1%".

I liked that.
 
I don't care if FreeBSD became mainstream
BUT if mainstream will make manufacturers program native drivers for their new hardware for FreeBSD
I say go mainstream
 
Crappy vendor provided drivers is not better then crappy OS provided drivers. If there's one thing to learn from Windows it's that.
Granted, good drivers from vendors are better then good drivers from OS, because they can focus on other things.
But let's turn it around, isn't it better to have OEM's like iXsystems and Freedom Technologies providing quality hardware for a quality OS?
 
Eponasoft said:
But in all reality, folks...would you really WANT FreeBSD to be a "mainstream" OS?

Yes. The more "mainstream", the more exposure and interest. The more interest, the more vendor support, and the more people hacking away at it and making improvements.

As long as those "half-wits" don't have commit bits, I don't really care about them. Any OS is going to have them.
 
Back
Top