Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
FreeBSD 9.0-Release Announcement (freebsd.org)
130 points by udp on Jan 12, 2012 | hide | past | favorite | 35 comments


FreeBSD AMIs are available for m1.large and larger EC2 instances: http://www.daemonology.net/freebsd-on-ec2/


That's great, good work!

How about using them on something EC2 compatible, outside of EC2? Are they available somewhere?


I'm not sure what you mean by "something EC2 compatible, outside of EC2"...


I'm sorry, I should've mentioned.

I was talking about deploying these images on EC2 compatible clouds - like openstack / Eucalyptus.

Some of us aren't customers of Amazon's EC2 service, so it would be nice if there was some sort of "how to" build these AMIs.

Other than that, I'm happy to see a new FreeBSD release.


Take a standard FreeBSD disk image. The changes I've had to make are to work around quirks in EC2, so I suspect most or all of them wouldn't be needed on other clouds.


Great, thank you!


It's awesome that they're including High Performance SSH (HPN-SSH) by default. This is a great project that deserves more attention: http://www.psc.edu/networking/projects/hpn-ssh/


This project does sound awesome and the release announcement is the first I'd heard of it.

However, there are a couple of issues. First, it appears that the patch set is no longer maintained. For the FreeBSD people out there, does inclusion in FreeBSD mean that that project has assumed the responsibility of porting the patches while they're needed, or should we just assume that FreeBSD 10 won't include the patches if a new SSH server comes out and the patches are not updated against it?

Also, this thread: http://lwn.net/Articles/377723 on LWN, containing input from an OpenSSH contributor (djm), claims that the HPN patches will not help until you are transferring at ~160Mbit/s @ 100ms lag and encourages normal users to benchmark the connection speed difference before assuming HPN will be faster for them. This is much higher than most anyone will get over the public internet, so should it be assumed that HPN is no longer necessary? Note also that thread is about OpenSSH 5.4. Have there been further improvements that have made the patchset even less necessary?


You raise some interesting points. 1) What makes it look like the project is no longer maintained? The latest patch release on the website is for 5.8p1 -- this is just 1 point beneath current (http://openbsd.org.ar/pub/OpenBSD/OpenSSH/portable/). 2) Great question. I would be interested in seeing the response this gets from the FBSD mailing list. 3) It is true that it is challenging (or rare) to get public internet speeds that will make HPN worthwhile, however I find HPN most useful on a LAN (typically within the same datacenter). I perform most of my data replication with rsync over HPN-SSH. It's multi-threaded and generally cuts CPU load over time. Thanks for linking to that thread over on LWN -- that was an interesting read.


I suppose "unmaintained" was incorrect, I probably should have said "not well maintained". I was going off of the notice on the HPN-SSH website at PSC: http://www.psc.edu/networking/projects/hpn-ssh/, which has a large bold header that repeatedly says that updates are "delayed" because all funding for the project is gone and there are no interested sponsors. So while not technically abandoned, the project is apparently "on the ropes" and further updates are questionable.

I usually use SSH on my local network too, but I usually max out the capacity of my disk in transfer speed anyway (gigabit network, but disk max read/write is ~30MB/s sustained), and there are definitely some fairly simple alternatives to local SSH usage if it's not fast enough in a particular situation (which I haven't encountered personally).

Of course, this doesn't mean HPN-SSH is useless, but its utility may manifest only in a small number of cases, making continued interest in maintenance or funding unlikely.


They also have clang which make this an interesting alternative to gcc for C and C++ developers.


Does this mean that all of the FreeBSD base is also compiled with clang ?


Yes.


All? Kernel + userland? Great!


It'll be far more interesting when lldb matures.


It's already extremely interesting with libclang, clang's static analyzer, clang's auto-completion, clang's error messages, clang's compile times, etc.

I am really looking forward to lldb though.


This release is a nice piece of kit for storage systems even if you're married to other OSes elsewhere.


If you're thinking of running FreeBSD on your desktop, checkout PC-BSD 9, it's a more desktop (dev workstation) friendly version of FreeBSD.

http://www.pcbsd.org/


Double-take >> The powerpc architecture now supports Sony Playstation 3

Is this actually possible now? or is the PS3 still locked to prevent actually installing this?


It supports only the old other-OS feature that Sony removed.


It's possible if you have a PS3 with the old firmware.


ZFS supports deduplication with this release.


Does it support installation to ZFS out of the box?


No. If you want ZFS root out of the box, you need to follow a guide like:

http://www.aisecure.net/2011/11/28/root-zfs-freebsd9/

It's roughly the same state as it was during FreeBSD8 except that some of the details are different due to the new installer.


hope linux can match up, esp on ZFS


I doubt ZFS is ever going to be worth using on Linux unless ZFS has its licence changed. My guess is that a new or existing file-system will implement ZFS's ideas and become usable long before that happens.


btrfs


Do the developers of btrfs recommend using it in production yet? last I heard they didn't feel it wasn't complete or stable enough yet.


Wikipedia lists it as being unstable. When it comes to file systems that's one thing I would not be gambling on.


No, they don't. Be wary. I tried to use btrfs in raid10 (data and metadata) mode for my latest array of large disks (4x 2TB WD Caviar Blacks); 70 GB in copying data from my old md array, I got a kernel OOPS. This was with very recent btrfs code -- Fedora 16 with a 3.1.x kernel and btrfs-progs-0.19-16.fc16. It's simply not ready.


Wow,hard to understand the FreeBSD installing menu, so ...hope I can become smarter.Should try some modified distribution


Erm, the new bsdinstall menu is about as simple as they could get. You have the option how to use your disk, with an "Automate" option to let the installer do it for you. Then there are four sets of "packages" that you may want to have installed, and then some options to define the configuration of your keyboard, location, timezone, et cetera.

In all, I think the bsdinstall is really a big step forward for FreeBSD.


There are no "distributions"


BSD: Berkeley Software Distribution

PC-BSD: based on FreeBSD


I'm still going to argue the point, there are no distributions in the sense of Linux. There's just "FreeBSD".

"BSD: Berkeley Software Distribution"

You're being pedantic here, again, I meant it in the context of linux distributions (that's what most people coming to FreeBSD are familiar with when they say things like, "I don't like the installer, I'll try another distribution!").

"PC-BSD: based on FreeBSD"

Right, this is the only thing I would say comes close to the general "idea" of linux style distributions - but I still think it is inaccurate to say that there are "distributions" of FreeBSD. PC-BSD even has its own kernel modifications in order to make it more desktop friendly - I would classify this as its own "BSD" IMHO (much as DragonFly was "based" on FreeBSD until they decided to completely fork it) even though it still relies heavily on the FreeBSD project.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: