FreeBSD 6.0 Released

Reyad Attiyat writes FreeBSD 6.0 is ready for release! New features, and there are lots, can be reviewed at the official site. One of the biggest and most anticipated features (mentioned before on Slashdot) is wireless support, which has been greatly improved upon. This includes support for a lot more cards, WAP support, and integration into the dhcpd client. This release comes only mere days off NetBSD's release and an OpenBSD release. Version 6.0 was intended to be released way back in August but due to a number of factors it had to be delayed till now. Aside from this major release the FreeBSD project has also had some major changes, including most recently a new logo and also a brand new website."

Comments

BSD is dead?

[eric2hill]

Looks like BSD lives to die another day :)

Variants.

[Anonymous Coward]

Is there a Dead-CD version of FreeBSD that I can boot on my machine, just to try out?

Real improvement over 5.x

[Just Some Guy]

I've been comfortable enough with the release candidates to upgrade my production servers from 5.x to 6 a while back. I really have nothing but good to say about it: it's faster, more stable, and more worthy of the FreeBSD name than 5 ever seemed to be.

Congratulations, Release Engineering team! You've turned out a great product.

And as a side note, we've seen big releases from each of the major BSDs within the last week. Dying, my foot.

Re:Real improvement over 5.x

[Anonymous Coward]

Please don't take this as criticism, because it's not meant in that spirit, but I really wish we'd stop even acknowledging the whole "dying" joke. The problem is, I've run into one stupid manager who, amazingly, bought it. Yes, oh yes, he was stupid, but somehow he got the job and I wish that I could say that he's the only guy I've ever met who was in a position he shouldn't have had.

Of course BSD isn't dying. Either that, or it's the longest death in history. Some Linux-based asswads -- and please understand that I'm not equating Linux usage with being an asswad, heck, I use Linux every day and I wouldn't want to incriminate myself; no, I'm just specifying asswads who somehow have found Linux -- just can't shut up. They're like the annoying nerd in the back of the room who has found an expert mix of geekness and insufferable rudeness that, for some reason, he mistakes for wit. Think "Malvin" in WarGames, and you get the idea. They're the guys who think they're experts because they've managed to install Debian and, oh, Gnome maybe, but they only use Synaptic for package management and they're afraid of cron even though they talk about how they use it all the time, even for personal tasks(!).

God.

They're *dying* to appear knowledgable, so they perpetuate this dying joke without ever really knowing what situations the BSDs are best suited for or the philosophies behind them. I mean, the BSD Web sites, have you seen them? That have all that *text*, I mean, God, you have to read and stuff.

FreeBSD has its issues, I'm aware of them. But I've used it for many, many moons now, and honestly, it rocks my world as a server system. I might use something else, but I'd have to have a really, really good reason. My mail servers run on fairly close-to-stock OpenBSD systems, and they're rock solid. Package management and upgrades are a breeze.

Bottom line, the BSDs make my job easy.

Re:Real improvement over 5.x

[jadavis]

Can you tell me a little more about any issues you had when upgrading? I just installed my first production FreeBSD box (using 5.4) a month ago.

So far I've had a great experience with FreeBSD. Before that my only experience was GNU/Linux (mostly on Debian).

Any useful resources you have would be nice too. I'm still relatively new to FreeBSD, and I'm not familiar with everything about it. If it's not in the handbook, I don't know it.

Re:Real improvement over 5.x

[Just Some Guy]

Do I need to recompile all the ports I installed against the 6.0 system?

You probably don't need to, but it's a good idea. Here's how I do it:

Use "find" to find and delete all files in /bin, /sbin, /usr/bin, /usr/sbin, /lib, /usr/lib, /usr/libexec, and /usr/include that are older than the day I upgraded the machine. At this point, I have a 99% "pure" base system, but almost all ports on the machine are temporarily broken. Note that FreeBSD 7 has this feature built-in.

Is portupgrade installed?

- If so: cd /usr/ports/lang/ruby18; make deinstall; makereinstall; make clean. This rebuilds ruby against the new system libraries so that portupgrade will work again.

- If not: cd /usr/ports/sysutils/portupgrade; make install; make clean to install it.

Upgrade all the ports in one swoop: run portupgrade -fa and walk away for a few minutes/hours/days until it's finished.

There might be trickier ways to do it, but that way's pretty much guaranteed to work. Regardless, this is an excellent time to join one of the FreeBSD mailing lists. Everybody's going through this together, and you're very likely to get some friendly, useful advice.

Re:Real improvement over 5.x

[toadlife]

Yep,

5.x to 6.x is not a hard upgrade at all. My desktop at home was running 5.1 when I first set it up, and it's gone from there to 5.2 > 5.4 > 6.0_BETA1 > 6.0_RC1 with no issues at all. As for performance gains, I did do a run of UNIXBENCH on 5.4 and compared it to 6.0_BETA and I saw higher scores accross the board. Though UNIXBENCH isn't exactly real-world stuff, it would indicate that overall performance is somewhat improved in 6.

Re:Real improvement over 5.x

[Just Some Guy]

First, there are two mostly separate steps: upgrading the base system, and then upgrading everything else. Use the process you mentioned of running cvsup, buildworld, etc. (but be sure that your stable-supfile has the line "*default release=cvs tag=RELENG_6", or else you'll end up with the most recent 5-STABLE system!).

You are also correct that the ports tree is unrelated to the release version. The base system in 6.x comes with a new utility, "portsnap", that you can use instead of cvsup to update your ports tree, but it's perfectly OK to ignore that for now and use cvsup as usual.

I'm puzzled as to how portupgrade is supposed to know I want everything fresh and shiny with the 6.0 kernel.

Basically, the portupgrade (and the underlying ports sytem itself) will build everything to run on your current system. Running 5.x? It builds 5.x binaries. Running 6.x? It builds 6.x binaries. So, you'll want to run upgrade the base system and be booted into it before you start upgrading your ports.

WAP? oh yeah...

[FlashBuster3000]

Uhm, i don't think that WAP (like html for mobiles) has something to do with wifi,
most probably the article means this sentence:
"In addition to architectural changes, it includes completed 802.11g, WPA [...]"
WPA = security standard (stay back from WEP, guys!)

Re:how does this affect OS X?

[wvitXpert]

Why should it affect it at all? Just because Apple used FreeBSD code in OSX doesn't mean that they care what changes FreeBSD makes later on.

Re:Torrent?

[Just Some Guy]

If you're upgrading a 5.x machine, the usual cvsup / build / install routine will give you a shiny new system. It's a much easier upgrade than from 4 to 5.

Re:Torrent?

[cperciva]

Anyone know if there's a torrent available?

Yes. The official release announcement (which for some reason wasn't linked in the story) has a link to the torrent files.

Re:how does this affect OS X?

[Overly Critical Guy]

Darwin eventually syncs with FreeBSD, which ends up in major OS X releases. You can be sure OS X Leopard will be synced with 6.0, as Panther was synced up to 5.0.

Re:6-STABLE?

[cperciva]

The 6-STABLE branch starts with 6.0-RELEASE. The 5-STABLE branch started with 5.3-RELEASE.

From what I've seen, 6.0-RELEASE is more stable than 5.3 or 5.4.

Oh man ....

[gstoddart]

I just burned my 5.4 discs so I could upgrade from 4.8 two days ago.

Bah.

Re:how does this affect OS X?

[nine-times]

Well, it's not directly based on FreeBSD. OSX runs on top of Darwin (a different open source project), which was originally based on FreeBSD, but with a different kernel, and other changes. The people maintaining Darwin may choose to bring some of the new things over, but there's no direct connection (afaik) between an improvement being released for FreeBSD and seeing that improvement show up in OSX.

FreeBSD Java(TM) Project

[KlaymenDK]

http://www.freebsd.org/java/ !

There are several ports, one of them being native. Google! :o)

Re:Java on BSD

[bcrowell]

You can supposedly build the compiler and runtime from source using the FreeBSD ports system. However, due to licensing issues, you have to download a bunch of the files by hand yourself -- when I tried it, it took up basically a whole afternoon of clicking around on Sun's web site, jumping through hoops, and downloading 100-Mb tarballs. After all that, it failed to compile :-)

Gcj is becoming more and more capable, however. I believe it's now possible to build OOo with gcj as your compiler for the java bits.

Re:no offense, but give it a rest

[pmike_bauer]

OSX has had wireless support for ten years.
Good grief. BSD was once the great, now its like a garage hobby, albeit at an expensive elite univeristy.

Impossible.
1. A beta of OSX wasn't even released until 1999.
2. Although FreeBSD can trace its roots to the 4.4BSD from Berkley, the univeristy has nothing to do with active FreeBSD developemnt today (unless students & profs do work on their own).

Re:Live-CD?

[FunWithKnives]

Alright, can't you google for your own info?

http://www.freesbie.org/ - Latest release is based on FreeBSD 5.3.

It's harder to find LiveCDs of Open/NetBSD, but you can create your own -

http://ezunix.org/modules.php?op=modload&name=Sect ions&file=index&req=viewarticle&artid=88&page=1

http://www.onlamp.com/pub/a/bsd/2005/07/14/openbsd _live.html

Gee golly, and all of that was on the first results page after searching for "Open/Net/FreeBSD LiveCD".. Whooda thunkit?

A new release of DFBSD.

[CyricZ]

Unfortunately, we probably won't see a new release of DragonFly BSD until after the new year.

http://leaf.dragonflybsd.org/mailarchive/kernel/20 05-10/msg00030.html

For those who are unaware, DragonFly BSD is a heavily modified continuation of FreeBSD 4.x. It is done by Matt Dillon and many others who are/were prime FreeBSD developers in the past, but disagreed with the current FreeBSD development path.

Re:A new release of DFBSD.

[CyricZ]

The Wikipedia article does a good job describing the reasons for the fork:

http://en.wikipedia.org/wiki/DragonflyBSD

Put simply, Matt Dillon was a FreeBSD contributor for many years. He was responsible for some of the most important and beneficial work, such as the virtual memory subsystem rewrite.

He did not feel that the direction of FreeBSD 5.x was a good one, and thus did the open source thing and forked his own kernel from FreeBSD 4.8.

DragonFly BSD is incorporating many features useful for multiprocessor systems. It includes a vastly improved kernel memory allocator and improved threading and messaging constructs, for instance.

While it is still under development now, it is poised to become the leading general-purpose desktop/workstation/server BSD of the future. It's partaking in the innovations that will be needed with the advent of multicore and multiprocessor systems.

Re:how does this affect OS X?

[bugg]

Well, yes and no. Darwin has a mach-based "microkernel" but there's only one thread running under it - the FreeBSD kernel. So while the VM was supplied by Mach, the entire process model, network stack, filesystem code, and system calls all came from FreeBSD. The suite of userland tools that came with Darwin were all also from FreeBSD. They are, or at least were- I'm not sure how actively they merged things, but if the OP is right then they've kept up- very closely related.

New logo on web site?

[bhirsch]

I thought the web site was going to reflect the new logo...

I386 Support Removed?

[HighOrbit]

From the Kernel Changes part of the release notes:

Support for 80386 processors (the I386_CPU kernel configuration option) has been removed. Users running this class of CPU should use FreeBSD 5.X or earlier.

I sorta find that astounding (not that I have a 386 around myself). Oh well, the world has moved on.

Vast performance improvements.

[CyricZ]

Earlier today a client came to me, requesting that a FreeBSD 6.0 demo box be set up as a potential replacement for their current OpenBSD mail server. Indeed, 6.0 may be the release we have all been waiting for. The performance is vastly improved, and the stability is fantastic.

We found that the server was able to process about 60% more mail when running FreeBSD 6.0, as compared to OpenBSD. That's not to suggest that OpenBSD is bad, but performance wise, FreeBSD has taken the lead. And that was without significant tuning, and running a GENERIC kernel.

I'm not certain yet if it is improvements in the network stack, the filesystem subsystem, or in the scheduling. It may be a combination of all three. Some more time will be needed to determine exactly where the benefits are coming from.

Re:Vast performance improvements.

[Just Some Guy]

That's not to suggest that OpenBSD is bad, but performance wise, FreeBSD has taken the lead.

Erm, has OpenBSD ever had the lead in performance? I really doubt it; that's not what it was designed for. All of those little niceties like ultra-paranoid memory protection, cryptographically random process IDs, etc. take resources. Basically, it's tuned for security and correctness with a nod toward performance, while FreeBSD emphasizes raw performance over watertight security.

That doesn't mean that FreeBSD has bad security, or that OpenBSD doesn't incorporate performance enhancements when they can safely do so. All of the BSDs are heavily cross-pollinated, and the best ideas tend to get broad support from all of them.

Still, it's pretty reasonable to say that OpenBSD is more secure and FreeBSD is faster. I wouldn't be the least surprised that FreeBSD can process more email or web hits, especially when you through SMP or HTT systems into the mix.

Re:Vast performance improvements.

[ipso_facto]

Performance on OpenBSD is important but it's not the primary focus. If someone put an OpenBSD mail server in place knowing performance was going to be critical, then they chose the wrong system. If they chose it with security as the most important criteria then the move to FreeBSD will mean that this area isn't as strong as it was with the old system.

We all want the best tool for the job but the poster seems to imply that in moving from OpenBSD to FreeBSD there's a win in the performance arena without any loss elsewhere. OpenBSD focusses on security. On top of the system-wide security features, the in-tree OpenBSD sendmail instance has lots of OBSD specific patches (http://www.openbsd.org/security.html). For example, when OpenBSD chose their MP implementation, they deliberately chose biglock because of it's (relative) simplicity. This is important because the OpenBSD codebase is actively audited. The fine grained locking in the FreeBSD MP implementation is obviously going to blow OpenBSD away but at the cost of simplicity (bugs and security issues are harder to find in complex code).

ULE scheduler?

[pmike_bauer]

The ULE scheduler has been fixed, but is it enabled by default?
Could anyone explain the benefits of the ULE vs. 4BSD schedulers?
Are there real performance benefits?
Thanks.

Re:how does this affect OS X?

[athakur999]

The core of OS X, Darwin, is already available from Apple:

http://developer.apple.com/darwin/

Re:6-STABLE?

[DigitalNate]

There were a lot of problems in the early 5.x releases, which led to a lot of criticizm from everyone who was used to the rock solid 4.x versions. If you browse through the release engineering section on the FreeBSD website http://www.freebsd.org/releng/index.html, you will find some articles discussing changes to the release engineering process that occured in response to the problems experienced during 5.x.

Re:FreeBSD 6.0 = good

[fak3r]

from dmesg:

atapci0: <VIA 82C686B UDMA100 controller> port 0x1f0-0x1f7,0x3f6,0x170-0x177,0x376,0xd000-0xd00f at device 7.1 on pci0
atapci0: Correcting VIA config for southbridge data corruption bug
ata0: <ATA channel 0> on atapci0
ata1: <ATA channel 1> on atapci0
ad0: 39266MB <IC35L040AVVN07 0 VA2OAF0C> at ata0-master UDMA100
ad1: 114473MB <WDC WD1200JB-00EVA0 15.05R15> at ata0-slave UDMA100

I haven't had any problems with this setup for months on pre-6.0, so I don't think it's an issue, check your BIOS and be sure it's set to AUTO...also, any advice on RAID? After a failure last week I really want to get a cheap RAID solution for redundancy on this server. heard of some good cards from LSI.

Re:memory management and scheduler?

[Anonymous Coward]

5.x didn't retreat from those goals, and as the releases progressed from 5.2 onwards, the code matured and became faster/more stable. 6.0 is arguably a refinement of that work: now it has seen a few years of deployment, the developers have significantly optimized it, applied the refined approach to more components (VFS), etc. Whereas 5.x was about introducing the new architecture, 6.x is all about making it blindingly fast and stable.

Re:how does this affect OS X?

[Guy Harris]

Darwin has a mach-based "microkernel" but there's only one thread running under it - the FreeBSD kernel.

Or, at least, kernel code based on BSD - but developed independently; it's not just FreeBSD lifted up and modified to plug into Mach. In Tiger, for example, the MP locking, and VFS layer, are significantly different from FreeBSD.

As for userland, the system library (called libSystem on Darwin/OS X, unlike the libc on most other UN*Xes) is based on BSD, but not identical to FreeBSD's - for example, a lot of the get*by* routines just communicate with lookupd in Darwin/OS X.

Are you sure about that?

[CyricZ]

You do realize that JDK 1.4 is not JDK 1.5, correct? There have been significant improvements.

I recently set up several servers running FreeBSD 5.4, the native JDK 1.5, and Tomcat 5.5.12. They work perfectly, each handling upwards of 9 to 10 million hits per day.

Re:6-STABLE?

[cperciva]

6 just came out, [so] how can one possibly proclaim that the newer release is even more stable? Enough time hasnt passed yet.

Not at all. I (and many other people) have been running 6.0-BETAs since mid-July, and 6.0-RC1 since early October. This isn't just a random snapshot of HEAD; the code which became 6.0-RELEASE was frozen apart from patches approved by the release engineering team for months leading up to the release.

Re:But are the ports ready?

[questionlp]

The upgrade from 4.x to the 5.x branch was a significant change, everything from a major compiler version upgrade, compatibility libraries, and other core files changed.

I have migrated 6 machines from 5.4-RELEASE to 6.0-RC1 without any problems with recompiling ports (including: Postfix, Perl, SpamAssassin, Python, Apache 2.0, PHP 5, PostgreSQL). You may want to go with the stock GENERIC kernel configuration file that comes with 6.0-RELEASE and trim it back down as some of the options have changed.

Also, check out the changes in /etc/make.conf variables.

Digging up grandma?

[GooberToo]

Isn't this announcement like digging up grandma and throwing her a birthday party?

For the humor impaired, this was a joke!

Glad to hear that BSD keeps on chugging along...Linux has a reputation for taking some excellent ideas and even implementation from the BSD guys....BSD improvements often translate into Linux improvements too.

Good job guys!

Re:how does this affect OS X?

[Fred_A]

The more I look around MacOS X, the more it seems to me it has as much relationship with a normal Unix as Plan9 has. So they are kind of related, but there also are lots of diferences.

The major one being the insane filesystem layout. A number of links to more orthodox locations helps but it still is pretty weird and finding something isn't easy. Expecially since the included search system only looks in $HOME, so enabling locate is a good first move...

For most users who will never see the system anyway, this is a non issue. For all the Unix people who grabbed a Mac because the Unix side was important (for me it was the best value I found for a Unix laptop), it's a bit of a letdown. Of course it's always fun to explore something new. I probably need to find some proper documentation though (OReilly has a site on Unix/MacOS).

Re:how does this affect OS X?

[Eil]

Actually, userland for OS X is primarily netbsd derived, not freebsd.

Wrong. The page you linked to mentions all 3 BSDs exactly once, never specifying which one in particular the userland was primarily derived from.

I'm more inclined to believe the following, straight from news articles and Apple's own documentation:

"Going forward, [Darwin] will track a stable version of FreeBSD, which is the more popular and traditionally x86-only version that claims about a million users worldwide..." (source)

"The Darwin kernel is based on FreeBSD and Mach 3.0 technologies..." (source)

"...the BSD portion of Mac OS X is primarily derived from FreeBSD..." source)

"Above the Mach layer, the BSD layer provides "OS personality" APIs and services. The BSD layer is based on the BSD kernel, primarily FreeBSD." (source)

"We should note, however, that apart from a few architectural differences (such as our use of the Mach kernel), we try to keep Darwin as compatible as possible with FreeBSD (our BSD reference platform)." (source)

"Integrated with Mach is a customized version of the BSD operating system (currently FreeBSD 5)." (source)

In fact, practically the only references I can find to NetBSD in Apple's Developer Connection are to the HISTORY sections in some of the man pages. Apple may have borrowed some from NetBSD, but the main BSD player in OS X is clearly FreeBSD.

Re:Cisco VPN Client

[routerguy666]

I have been using this for quite some time now: [12:49pm exs@kirov /usr/ports/security/vpnd]> more ../vpnc/pkg-descr VPNC - Client for Cisco 3000 VPN Concentrator[12:49pm exs@kirov /usr/ports/security/vpnd]> more ../vpnc/pkg-descr VPNC - Client for Cisco 3000 VPN Concentrator A VPN client compatible with Cisco's EasyVPN equipment. Supports IPSec (ESP) with Mode Configuration and Xauth. Supports only shared-secret IPSec authentication, 3DES, MD5, and IP tunneling. It runs entirely in userspace. A VPN client compatible with Cisco's EasyVPN equipment. Supports IPSec (ESP) with Mode Configuration and Xauth. Supports only shared-secret IPSec authentication, 3DES, MD5, and IP tunneling. It runs entirely in userspace.

Re:Live-CD?

[misleb]

What do you think you are going to learn from a Live CD of FreeBSD? Whether or not it supports your hardware? I assume you've run Linux or some kind of unix variant. It'll have a shell and maybe a desktop like KDE or GNOME. What's to see? You can get that on just about any Unixy system. IMO, you don't really know what an OS or distribution is like until you have to actually manage a box.

Unless, of course, you've never run a unix-like system before. Then by all means, try Freesbie.

-matthew

Re:From what I have heard of FreeBSD

[Wonko42]

FreeBSD is ridiculously easy to use as Unix-like operating systems go. The installer is friendly and efficient, and generally (in my opinion) much easier and much faster than the installer for any Linux distribution I've ever used. It's entirely possible to go from a bare system to a working FreeBSD install with a generic kernel in fifteen minutes or less.

In addition, the ports system makes installing software a snap, and the online FreeBSD Handbook and FreeBSD FAQ are very well-written and kept up to date if you have any questions.

Re:Shit...

[Billly Gates]

I have been becoming more and more of a fan of netbsd's pkg's. DrangonFLY recently switched to it with Matt Dillion claiming the problems with the ports is that things change and break. Most DBSD users just used FBSD ports with some patch files. Shudder

I am looking into NetBSD after the fallout with FBSD 5.x but I am willing to try again with 6.x. Pkg's are also available for FreeBSD users from www.netbsd.org and you can compile them from source as well. It takes care of alot of dependancy problems.

>386BSD

[tverbeek]

This is something of a landmark, kind of like when Mac OS stopped supporting 68000 processors. These are the CPUs that these OSes were built for, and whose consistent feature set made it possible to engineer software to run on "any" computer of that kind. The idea of 386BSD not running on a 386 is a bit... eye-opening.

Re:From what I have heard of FreeBSD

[Eil]

Is that it's hard to use.

Rubbish. It's BSD, which means it's a Unix derivative. If you're familiar with one, you're familiar with them all. There are some differences, of course, but a skilled administrator or gifted hacker could bring themselves up to speed in an afternoon. If by "hard to use", you mean "there's no pointy-clicky administrative interface" then perhaps. However, most system administrators who have to deal with production systems all day long (FreeBSD's target audience) don't want any of that anyway.

I'm not sure what kind of environment FreeBSD is deisgned for (servers, desktops?)

FreeBSD is almost entirely geared toward servers. There are various movements and projects underway to help bring FreeBSD closer to the desktop, but it's nowhere near as strong as Linux in that arena. Linux is a good general-purpose Unix-like OS, but FreeBSD is typically regarded as a more solid server.

but the fact that I don't even know that says a lot

Forgive me for being blunt, but it only says that you didn't even bother looking into it. All of what I've stated here is fairly common knowledge to those who care enough to find out for themselves. We live in the age of the Internet. If you're curious about FreeBSD or any other operating system, go read up on it, don't just sit around to hear about it.

Re:Live-CD?

[mnmn]

LiveCDs make for quick evaluations. I dont have the partition for openbsd, so I installed it in vmware to check out the ospfd, how to start it and what can it do. I'll continue to spend time on it, to see if I can port ospfd to other unixen, and even cygwin, my eventual goal.

Now if I had a livecd, I'd use that instead to get full CPU on my side.

Now before you call me a newbie, search my name on google, and maybe visit me sometimes. I have stacks of sparc, hp, rs6000 and alpha machines in my room to play with (just enjoy trying out unixen and porting stuff in between). But I highly value liveCD, since you can download the latest version, plop it into any machine and off you go for the more basic stuff.

Try knoppix sometime. "knoppix 2" gives you just the command line.

Re:Java on BSD

[EvilTwinkie]

Ive had to build java from the ports several times, and while yes its fairly painful the first time, with the login and accounts with Sun, its always worked like a charm. Outside of the compile time, and the hoop jumping, its stable, and works perfectly. I dont care much for precompiled binaries as thats the lame way of building a box. Personally, its always from source, and from the ground up. I've gone as far as to roll my flavor of bsd which is *exactly* what I want. It starts at 5MB... Shame Sun has to be pissy and moany...

Re:Shit...

[NuShrike]

At least cvsup rebuild to RELENG_6 from 5.4 is much easier and painfree than moving from 4.11 to 5.x.