Submitted by FreeBSD 6.2, one of the most popular versions of the free BSD Unix operating system, is out with new features and updates. It plugs holes and, in addition to the usual route of installing directly to a hard drive this time around, offers a LiveCD that can be used to rescue a broken system. Developers on the free version of the BSD Unix operating system hope it will help build momentum for what is arguably the most popular BSD variant in use.
I’m currently downloading the first cd with BitTorrent, and I’m really excited about this. I’m taking a look at the installation handbook (a damn good doc, I must say) and I just joined the spanish mailing list. This will be my first FreeBSD installation, I feel like a kid with a new toy xD
Enjoy
For those who don’t have a spare box..
http://www.thoughtpolice.co.uk/vmware/#freebsd6.2
Thank you very much!
I am torrenting it now.
I’m setting up a system right now. I actually installed from the 6.1 CDs I had downloaded not too long ago. I didn’t want to re-download 6.2. Of course now I’m waiting on everything to update now. Maybe downloading the new CD would have been faster.
I hope to switch over to FreeBSD from Suse (which I’m getting sick of). Though I might go with PC-BSD for a little more simplicity (at least for my desktop/laptop).
As long as you’re aware of what you’re getting yourself into, you should be okay; It takes several hours to set up and customize a desktop system, but many FreeBSD users feel it’s well worth it. For those who don’t, there is of course PC-BSD (who have a new release out now) and DesktopBSD (who I believe are about to issue a release candidate for their latest version soon).
I’ve built up several pages of notes over the years on how to set up and customise a fresh FreeBSD install to my liking. I’m sure you’ll end up doing the same (if you stick with it). In the meantime, a good website (particularly for newcomers) with similar tips is here:
http://www.engr.colostate.edu/~reinholz/freebsd/
And don’t forget:
http://www.bsdforums.org/forums/index.php?
which contains a wealth of information.
Have fun!
If you are new to FreeBSD it may last serveral hours, but I do it in one hour from source. Of course my desktop isn’t KDE or Gnome
But last not least, you don’t have to install FreeBSD all the time
As long as you’re aware of what you’re getting yourself into, you should be okay; It takes several hours to set up and customize a desktop system, but many FreeBSD users feel it’s well worth it. For those who don’t, there is of course PC-BSD (who have a new release out now) and DesktopBSD (who I believe are about to issue a release candidate for their latest version soon).
Oh, I know what I’m getting into, don’t worry
I’ve already installed PC-BSD before, but I didn’t really like the PBI idea… I might give DesktopBSD a try as soon as it’s updated to 6.2, but I want to try The Real Thing. It can’t be much more difficult to set up than Slackware, Debian or Gentoo.
And thank you for the links!
Hi all,
I’ve heard many people saying that the project is somewhat stalled, there isn’t many new features, there are mainstream technology that doesn’t exist in Freebsd and so.
The freebsd developers are putting an enormous effort in FreeBSD 7. There are many things that are being re-written, many new features which include iSCSI (this is what most people complaint), ZFS and XFS, Dtrace, file system journaling in the GEOM framework, support for new archs (sun4v and arm), improvements in the current development of sched_ule and sched_core, kernel threading systems, improvements in the sound subsystem, new drivers are being commited (including the famous 3945bg), cutting edge security features, and much much much more.
While some of this is going to be backported I’d say that FreeBSD 7 will be the true milestone for the project.
So people, let’s help out, even the basic user can test stuff and file pr’s.
Cheers
I’ve heard many people saying that the project is somewhat stalled, there isn’t many new features, there are mainstream technology that doesn’t exist in Freebsd and so.
Yes, their unmature opinions are very misleading.
Going that way projects like Slackware or Debian can be considered as dead
I’ve heard many people saying that the project is somewhat stalled, there isn’t many new features, there are mainstream technology that doesn’t exist in Freebsd and so.
A native Flash implementation is about the only thing mainstream that doesn’t exist. Java? Check and pretty much everything else which works on Linux works on FreeBSD. I run across the occasional close-source app which won’t work, but I wouldn’t call any of them main stream.
Yes, FreeBSD 7 will rock!
Hi
In my opinion the only mainstream technologiy that lacks in freebsd is iSCSI and, but that is being addresses.
There are other little things that I think are important but don’t oddly exist. For example a decent port channel implementation. ng_one2many only supports round robin (who uses that?) method. It’s useless basicly.
The important thing is getting users and companies to aid the project somehow. The developers can’t do everything
Cheers
I have to say that the only reason I don’t primarily use FreeBSD is it’s lack of 3D support for my video card. Under Linux, my ATI x800 XL is supported by the open source drivers, has full AIGLX support, and works with beryl and compiz. Under FreeBSD, 3D won’t work at all due to problems with the pcigart support for ATI cards. On another machine, with an AGP x700, 3D works, but AIGLX currently locks up the system when launching X (and AIGLX locks up the on systems using the intel driver when quitting out of X).
There is plenty of activity on getting Xorg 7.2 into the ports tree, but there seems to be very little work on getting these technical hurdles fixed, unfortunately.
A good page to read about FreeBSD 7 is http://ivoras.sharanet.org/freebsd/freebsd7.html … I fear that 7, like 3 and 5, will be a step child to 6->8.
That said, I am looking forward to any thing that the DFBSD project provides for FreeBSD down the line.
>>A good page to read about FreeBSD is
>>http://ivoras.sharanet.org/freebsd/freebsd7.html … I >>fear that 7, like 3 and 5, will be a step child to 6->8.
Funny that you say that, I read somebody else’s opinion that 7 will be more of a stabilizing release since 6 didn’t do enough of that for 5.
Anyway comparing to some of the changes that went into 5 (I can’s speak for 3, I wasn’t aware/paying attention back then) 7 doesn’t even count as invasive enough to need a whole major number release to stabilize.
>> 7 doesn’t even count as invasive enough to need a
>> whole major number release to stabilize.
As I understand it major release numbers amount to a bump in shared library versions. For example 6-STABLE refers to libs/kernel interface stability for the lifetime of 6.x, while moving an application to 7.x would need to be recompiled against the new system.
Edited 2007-01-19 21:02
“For example 6-STABLE refers to libs/kernel interface stability for the lifetime of 6.x, while moving an application to 7.x would need to be recompiled against the new system. “
If I remember correctly, the ports misc/compat[2345]x allow you to avoid such a recompile run.
… and now with latest desktop goodies: soon, xorg 7.2 will hit ports, along with beryl and friends. Tested it with latest nvidia driver, and works fine (cube, ripples, wobblywindows and all).
About flash: flash 9 was just added to ports. If you really need flash, install linux-firefox: it is just as up to date as native firefox, without any performance drawbacks. With linux-mplayer-plugin (which uses native mplayer) and realplayer, it gives you a complete web experience (and unlike the attempts to make linux flashplugins work with native versions of the browsers, this combo is pretty stable).
Now only if I could provide some screenshots of beryl! Unfortunately, ksnapshot outputs a somewhat garbled image of the cube:
ftp://hatvani.unideb.hu/personal/screenshots/beryl/cube1.png
.. but actually, it looks fine. And it is surprisingly useful (as in being a usable replacement for desktop pagers).
I tried to write BSD on the screen with water effects in this one:
ftp://hatvani.unideb.hu/personal/screenshots/beryl/watereffect1.pn…
… though it didn’t came out very readably
FreeBSD 7 will also be great! http://ivoras.sharanet.org/freebsd/freebsd7.html (shed_core and shed_ule – I’ve been using the ULE scheduler since 5.1, and I never had a problem with it, although it is deemed unstable in its current incarnation. ULE 2.0 is expected in 7, and I only mention it because for desktops, it gives you much better performance and interactivity even if your machine is heavily loaded). Multimedia support (for Intel HDA, low latency patches, video capture cards) is improving at a fast pace, while we’ll see a brand new USB stack in 7 – things are looking up for FreeBSD
)
Binary package management. True pkg_add will do the initial install but maintanance of the system on an ongoing basis is painful involving portupgrade. One can choose the binary only option, but then I have had to deal with many dependency issues later on. Source build of large systems and updates are quite time consuming – say openoffice or Gnome. Having said that, I still prefer FreeBSD to others. Have tries Ubuntu, Fedora, etc. Nothing like FreeBSD. Stable, fast and easy. Now only if we could get the Debian’s apt-get system ported to it
Agreed. I really enjoy building from Ports, but I’d love to have a package repository with up-to-date packages, especially regarding huge projects like KDE and GNOME.
I understand that the FreeBSD project simply lacks the resources to provide such a repository, so I wondered if the community could get together and write a @HOME-like client that silently uses spare CPU cycles to build packages in the background and send them to central servers when finished.
Don’t know if that’s feasible, I’m not a programmer, and I guess there would be serious security and administrative questions, but such a client would simply rock.
Even if you have many machines, as long as you share similar architecture and not every machine requiring different softwares and hardware, you can simply build on 1 machine, which can be a dedicated ports build machine, then distribute the prebuilt packages to the rest of the farm.
FreeBSD or Gentoo Linux has this nice mechanism to get you your own made ports as packages.
I just can’t go back to binary distribution these days, when the real drawback is that you can’t choose the feature of the application as well as dependency of it (Though x11 dependency on binary packages are usually sorted out as no-x11 packages, but that’s a bit of a dirty hack). As I always say, installing more, decreases security.
Check out the Debian kfreeBSD port, it is advertised as immature, but should ship real soon now, with the next stable release
http://www.us.debian.org/ports/#nonlinux
s/package mgt/3-page config/usage guide “as” pkg-descr/g
but seriously…
*********************************************
“pkg add will do the initial install”
***********************************************
many people only pkg-add a few of the thousands of ports they will use…
*************************************************
“a painful process involving portupgrade”
*************************************************
portupgrade is very often suggested, however very
useful alternatives exist:
portmanager
portmaster
others…
…………………………………..
OTOH after finding several ports which failed to build with
“make build” but completed fine with
“make patch && make configure && make build”
I put together a [something].sh (living in /usr/ports/)
(run like #port## source ../../[something.sh )
which does the manual-sequence-with-sound-notifications;
just this morning I updated it to include timestamps so
the next time I build the port i’ll know in advance about
how long it will take…
……………………
lots more to discuss but running late today… sorry…
………………….
Is there an explanation anywhere of what each disc is for?
I’m not entirely sure, but the first CD appears to be to install the primary system while the second CD is packages from the ports collections. When installing I chose to not install from the content on CD 2 and will download more current versions from the on-line repositories.
FreeBSD 4.0+ has always had a live fixit CD option. Prior to FreeBSD 6.0, it was CD2. With FreeBSD 6.0, CD1 became a hybrid install/fixit CD. Just select the Fixit option in the install menu.
Time to perform the upgrade dance on my servers
I wish I could say I liked it, but my experience with FreeBSD 6.2 wasn’t so great. On my 3-year-old laptop (IBM X31 ThinkPad), the last version that worked reasonably well was 4.10. The whole version 5.x series was essentially unbootable. Now with version 6.x, I can boot at last, but the OS crashes hard during shutdown, required me to pull the plug and remove the laptop battery. There are some other annoyances – the Centrino Wireless is not supported by default, though I understand you can get it to work after playing with ndiswrapper, which I haven’t yet done. Then there is the fact that my USB mouse won’t work with it, though thankfully my USB keyboard does (in 5.x it did not). Then there is the notorious geometry bug, which bites anyone trying to share a hard disk with FreeBSD and other OSs (no problem if you give the entire hard disk to FreeBSD).
All of which makes me ask if it’s worth the bother. It’s a pity, I wanted to like FreeBSD, but it’s going to need to become more hardware friendly. I imagine that with a generic beige box with FreeBSD as the only OS, there would be no problem, but that just doesn’t match the hardware that I have.
I’ve got ThinkPad R50e. I can use both, USB mouse and mouse “button” in the keyboard which is described as PS/2 mouse. I haven’t done anything special to make it work….
The only thing that doss not work is Conexant modem.
DG
There is a sort of live CD in 6.1., too. I know, because I have done something stupid, and had to boot from CD to repair the system.
DG
The primary reason why binary packages of Ports are difficult is that an awful lot of them are configurable in many different ways. To do binary packages properly, you’d need to either make some fairly scathing decisions about the packages and alienate a good proportion of your potential user-base, or produce a huge multitude of packages.
Simple example: many packages can build against either Heimdal or MIT Kerberos. The reasons for choosing one or the other generally lie with the system administrator (it’s a decision that the FreeBSD project can’t necessarily make for you). Which do packages get configured for? If it’s one, what about the users who’ve got a setup revolving around the other? What about all of the other similar choices (OpenSSL versus GNUTLS, for example)?
The Ports system is very flexible, but this is unfortunately the price that flexibility comes at. It’s by no means a show-stopper (e.g., if configuration is the default, fetch binary packages, otherwise build from source), and I could’ve sworn that portupgrade had support for binary packages, but it was pretty shoddy (there was no distinction between a package built with one set of options and the same package built with a different set, and you could easily end up with much older versions of software installed by expressing a preference for binary packages instead of building from source).
In short, it is possible, it just needs very careful management, and I’m not sure anybody’s brought a good solution to the table just yet.
Edited 2007-01-19 00:37
There is an easy way to make people happy even if you provide preinstalled binaries: let people install the binary, then whilst they are using that they can install the port with all the features they want; the package can then be uninstalled on successful installation of the port. This is the way I use Gentoo.
The one thing that Gentoo really lacks in this area IMO is precompiled KDE and GNOME binaries.
Edited 2007-01-19 01:03
I like Gentoo, and still like Gentoo and I like FreeBSD too, of how they can be built from source, as how I said above this, but what Gentoo lacks, (apart from your time consuming GNOME compile, but I really think if one wants a desktop system, they can simply use binary distro where there are PLENTY like Ubuntu, Fedora and whatnot, because seriously, desktop users don’t really care if their underlying system has mysql support or pgsql support on their postfix and so on by accepting the distributors binary package options), is their python scripting capability…
I don’t know who initially wrote that Gentoo base system, but even if I haven’t looked at the source of those ebuild system and emerge, they must be far from optimized, as in programatically and logically.
Gentoo takes like forever to make package catalog updated, when compared to like Debian or FreeBSD, I feel super fast. As well as Gentoo’s package searching system takes quite a long time to display especially making it to look for string in package description as well.
So, Gentoo, if need to be considered in serious scenes, they better start making things look like an enterprise level by start optimizing those scripts, but not as hobby hacked scripts. The core idea of Gentoo is superb, I ‘rarely’ see portage complain about unresolvable dependency like Fedora keeps telling me and everything starts compiling as I just tell it to compile the application’s name which I want to use (’emerge gnome’ on a fresh installed DOS-like situation will compile everything up, but takes whole night), which is the best package management system available yet to date, you don’t have to do ‘whereis gnome’ to find the directory structure as in FreeBSD, but Gentoo, just the underlying scripts aren’t as professional as it should be…
I agree – but that’s not necessarily a barrier. SuSE, after all, positions itself as an Enterprise solution, and YaST is SLOW. A lot of people complained about the speed of software installation in 10.1 but the problem was not that it was slow – it was that it was EVEN SLOWER.
Of course in SuSE’s case the problem even extends to other features of YaST like changing the wireless configuration.
That’s the way I use FreeBSD: do a minimal install from CD, use pkg_add -r to install the apps I need, then (as time permits) use portmaster to upgrade to the latest versions available in the ports tree (updated via portsnap).
Packages in FreeBSD are great for that. They give you a quick and easy way to get a system up and running so that it is usable right away. Then you use the ports tree to get the latest and greatest, or to get any missing features.
Well they could take a bit from OpenBSD’d ports tree. Each package is actually named with the options it was built with. Say for example I build vim with perl support, ruby support and no X11, it would come out with something like this.
vim-perl-ruby-no_x11.version.tgz
good article liked it,but it forget to mention freebsd’s 6.2 known problems that have pop up within three days freebsd 6.2 was released,only three days and already there are problems. what if a bank or hospital is using 6.2, and one of these bugs causes them to lose data? that would not be a good thing. I personally think freebsd is trying to advance the OS to fast,and not auditing the code for bugs that much. and it also seems that the BETA and RC CDs are not helping at all, OpenBSD does not use them and still their os is more stable,with much, much less problems. i wonder in 6 mo how many problems will be discoverd in 6.2. i think they should forget about freebsd 7.x and fix freebsd 6.x first. I’m happy i’m using OpenBSD as my main OS.
Known Problems and Solutions
[20070116] As mentioned in the release notes, the arcmsr(4) driver was updated for FreeBSD 6.2-RELEASE. Some isolated crash problems were noted after this update, although the driver in general tends to work quite well for many testers, and it seems to be “better” in general in terms of functionality and supported hardware. Users having problems may wish to consider backing out the driver update or installing a version of this driver directly from Areca. Further investigation is continuing into these issues.
[20070116] FreeBSD 6.2-RELEASE has been diagnosed with kernel bug related to UNIX-domain sockets. A patch for this problem has been committed to HEAD and RELENG_6, and is a candidate for an errata patch to RELENG_6_2 after further testing.
[20070116] Systems with very heavy network activity have been observed to have some problems with the kernel memory allocator. Symptoms are processes that get stuck in zonelimit state, or system livelocks. One partial workaround for this problem is to add the following line to /boot/loader.conf and reboot:
kern.ipc.nmbclusters=”0″
Other workarounds have been discussed on the freebsd-current@ and freebsd-stable@ mailing lists.
“good article liked it,but it forget to mention freebsd’s 6.2 known problems that have pop up within three days freebsd 6.2 was released,only three days and already there are problems. what if a bank or hospital is using 6.2, and one of these bugs causes them to lose data?”
Uh? I do not understand your point.
Did you already see a bank or an hospital installing a new operating system only some days after it was released?
There are always some tests and checks before that!
My enterprise has to renews its windows computers, do you think Vista will be on there? Well, no, not before at least a whole year!
It is the same thing with FreeBSD. The enterprise world is very cautious. So when data and system management are critical like in hospitals, banks, or whatever, they do not play with new shining toys. They wait to be sure the toys is harmless, even if a new better one is out in the meanwhile.
He means, FreeBSD better audit a bit more before a new release to excite users.
We know banks and hospitals don’t use FreeBSD on their critical computing requirement… but that doesn’t mean FreeBSD wants to stay cheap on quality by making several fixes within a few days of release.
It doesn’t mean, don’t make a fix, but they should be done before the release, because keeping it 1 week longer from release should have had some bugs fixed than making people moan about how new releases are unreliable.
We know enterprise never just put OS that was officially released before testing it themselves… but still it doesn’t mean OS should rely on users to test everything out…like forcing all of us as beta testers.
“We know enterprise never just put OS that was officially released before testing it themselves… but still it doesn’t mean OS should rely on users to test everything out…like forcing all of us as beta testers.”
That’s exactly what enterprise users do for production systems all the time as a standard practice — they test it before going live, no matter what O/S. This (a few bugs after release) doesn’t affect those types of users. IMO it’s some Linux distributions that put their users into beta testing on final releases *cough fedora cough*.
The original post wasn’t about users though, it was about banks & other high availability demanding services having to work through bugs after a final release. Banks, etc, are probably still evaluating one year old (and older) releases for production use, let alone a brand new bleeding edge 1-2 week old release. This is an absurd argument that new releases should be %100 bug free. No operating system would be released in years with that kind of methodology. There are other ways to deal with this issue, namely patches for known issues for known O/S releases, as long as they patches work.
Edited 2007-01-20 01:00
The original post apparently didn’t talk banks and space shuttle actually use freebsd, nor asking freebsd to be 100% bug free on launch, but finding just several bugs already right after a launch sounded as if they didn’t take enough care to make a stable release but rushed the release out, as a new release is about fixing bug after feature freeze.
True enough sithgunner. So are these bugs show stoppers? If so then yes it’s not a ‘good thing’ but if they are obscure bugs then it’s not a big deal.
Three bugs in three days is nothing. I would like you to point me to an OS that has not had a single bug in it. It is just plain impossible to test all code for all situations using all possible hardware combinations with all hardware revisions. I think the vast majority of bugs in 6.2 have been stomped out really well. For the rest of the bugs, people will have to just install FreeBSD on their test machine and verify everything is okay. Chances are that 99.999% of the people will run without a single problem. The rest will just have to work with the mailing lists to isolate the problem and get it fixed.
Hi,
Was anybody successful in getting the Nvidia 9xxx drivers isntalled?. I get an error in the make install steps. It doesn’t seem to find where the X11 directories are located.
Are you installing the driver from the ports tree?
Adam
If you are installing NVidia package, not from ports, try the following:
make all install X11BASE=/usr/X11R6
i am using this OS right now at home – a better configuration method for X is essential, and perhaps some better autodetection of sound cards. getting the user quickly to a desktop is one place where modern linux distros have bsd beat, but this is something that is doable and in 2007, should be there for freebsd too.
anecdotally i see freebsd6.2 being faster on the same hardware than debian etch.
Browser: ELinks/0.11.1-1.2-debian (textmode; Linux 2.6.18-3-686 i686; 83×25-3)
I found the following guide very useful: http://www.vtbsd.net/quickstart.html
I have installed FreeBSD 6.2 on a white box PC upon which Slackware 11.0 ran without incident. Both the 6.1 and the 6.2 release of FreeBSD have a bad savage chipset driver. Screen freezes up whenever you log out and all of the options are not available.
It appears that the development team forgot about this bug that was filed a short month after 6.1 came out.
Where is the quality control here? Linux has the edge in this department I guess.
Anyone know of a fix? Right now I am running FreeBSD 6.2 with the generic frame buffer albeit without scroll mouse support.
Thoughts anyone?
My servers and laptops run faster with FreeBSD 6.2, thank you for all this great hard work!