Guest post by For all of you using FreeBSD and ZFS, Kip Macy (kmacy) and Pawel Jakub Dawidek (pjd) merged ZFS Version 13 into FreeBSD 7-STABLE. Here is a breakdown of some of the new features: kmem now goes up to 512GB so arc is now limited by physmem, the arc now experiences backpressure from the vm (which can be too much – but this allows ZFS to work without any tunables on amd64), L2ARC Level 2 cache for ZFS which allows you to use additional disks for cache, and more.
And performance is still atrocious.
I’m having some issues where my fileserver keeps restarting on its own every 5-17 days (it’s inconsistent) and running OpenSolaris ZFS 13..
Looks like now I can run my array off FreeBSD while testing to see if it’s hardware or software related!
Thanks OSNEWS!
if you mix BSD and CDDL code what licence is the derived code subject too?
surely its CDDL. so will it be renamed to FreeCDDL?
or maybe the result is BSD. in that case ‘woot’. because BSD+GPL = GPL, so it can go in the linux kernel 🙂
or do you just get something undistributable? like when you mix GPL with proprietary?
Under the OpenSolaris licencing FAQ there is the following which clears things up:
Thus I assume that GPL/LGPL isn’t a file based licence (someone correct me if I am wrong).
Of course it is in CDDL. What made you think otherwise? Besides, not everyone is crazy enough to name their operating systems as X/Y.
I also wonder about these demands that the codebase of OpenSolaris should be GPL-compatible. Not because it would be a good thing for OpenSolaris, but because that way Linux people could exploit it? Makes me sad to even think about this possibility.
They’re not demands, although some people like to think they are. The fact is that if you’re going to say that you have an open source project then it seems reasonable to see some source code actually going into it with a license that actually has a point to it to facilitate that process.
Keep being sad because I doubt whether anyone in the Linux world cares about the OpenSolaris codebase. Linus himself has said as much from a kernel perspective. Solaris has nowhere near the depth or quality of device drivers that Linux has, especially on x86 which is where it all started for Linux and where it all started to go wrong for Solaris. Solaris’s userspace is also gradually being forced kicking and screaming into the 21st century with you guessed it, much of the (L)GPLed userspace software you’ll find as standard on all Linux distros.
Really, it’s sad to think of how irrelevant much of OpenSolaris’s codebase is now that large parts of it have been ‘open sourced’………eventually.
It was mentioned
“… I doubt whether anyone in the Linux world cares about the OpenSolaris codebase.”
Well, regarding the state of these OSes, we have a “newbie” operating system (i.e. Linux) that is not UNIX (i.e. is not able to adhere to a UNIX (i.e. properly designed) specification; where’s IBM/HP to help with this ?) and had been developed from someone’s bedroom versus a “beast” operating system (Solaris) developed in a commercial environment which has a proven commercial track record satisfying real-WORLD constraints and real-world warranties.
If users/developers of this “newbie” operating system have a hard time respecting the “beast” operating system for what it is then it’s their problem and not a problem for (Open)Solaris user/developer/admin.
It was mentioned:
“… I doubt whether anyone in the Linux world cares about the OpenSolaris codebase. Linus himself has said as much from a kernel perspective.”
Open-sourcing of OpenSolaris ensures the existence and advance of multiple OpenSolaris-based distributions. It appears a bit brave to think that Linux is the be-all/end-all in the operating system front and thinking that anything being open-sourced was intentional for the benefit of Linux. Actually, if Sun Microsystems were interested in Linux very-easily scalping Solaris-based technologies then OpenSolaris code would have been released using the viral GPL licence.
Do you really believe a non-standardised, hacked-away/unbridled operating system has serious use in life-support medical equipment, nuclear power station management, or operating system for multi-million dollar scientific equipment ?
Last time I checked (back in my doctoral work in late 1990’s) this was not the case and for obvious reasons.
Although it has it’s place, Linux has it’s serious “cracks”.
It was said:
“Solaris has nowhere near the depth or quality of device drivers that Linux has …”
If it came down to device drivers then we all would be using Microsoft’s operating systems and there are reasons why alternative (e.g. UNIX/UNIX-based) operating systems are preferred over Microsoft’s operating systems. As with Theo de Raadt’s view (Re: OpenBSD) we all realize that if the hardware companies were more forthcoming with hardware documentation then device driver support would be much better for many operating systems.
It was said:
” .. especially on x86 which is where it all started for Linux and where it all started to go wrong for Solaris.”
Really.
Why is it that OpenSolaris on my Opteron-based HP xw9300 is superb for my user/development needs ?
Why is it that on installing FreeBSD, Linux, and OpenSolaris distributions on a new x86-based system for an existing/happy Linux user friend of mine it had caused my friend after about a week of testing to realise that OpenSolaris would be his preferred non-“Windows” platform ?
It was mentioned …
“Solaris’s userspace is also gradually being forced kicking and screaming into the 21st century with you guessed it, much of the (L)GPLed userspace software you’ll find as standard on all Linux distros.”
I hear no complaints from my OpenSolaris system whenever it runs (L)GPLed userspace software that Linux also runs.
It was said:
“… it’s sad to think of how irrelevant much of OpenSolaris’s codebase is now that large parts of it have been ‘open sourced’………eventually”
So, is Linux’ code largely irrelevant because it was open-sourced ?
Still, why is it that cloning of Sun technologies like {ZFS, DTrace} got a junp-start after these technologies were open-sourced ?
you are very off-topic, and one of the reasons that made me move over to BSD side from Linux(or GPL) was zealots like you.
You use BSD without using a single bit of GPLed software? Is that even possible?
Personally, I like BSD-like licenses more, but the Linux kernel has broader hardware support than the free BSDs — especially for laptops (Mac OS X is a different story, though).
You use Linux without using a single bit of BSDed software? Is that even possible?
The point being?
Just wondering why on a site called OSNews, we have articles in the front page about an Ex Microsoft employee, about a web broser, about a DBMS, about some law, a review of a game, etc and a news piece about an OS is on the Page 2…
— EDIT: Corrected a typo
Edited 2009-05-22 14:40 UTC
I think it is about popularity.
You know, more hits and clicks for OSnews from every nonsense Linux article rather than from an interesting article about “alternative” or “exotic” operating systems.
Sounds harsh but was meant as a gentle criticism.
I think you are right
But an article like this getting to front page wouldn’t harm OSNews’s earnings that much in my opionion…
But anyway this is just my opinion and i really like OSNews. I intend this to be constructive criticism.
IIRC front page news is for original content only. Page 2 is for link to other sites with short summaries.
submit long news and it’ll be in front page (iirc ‘front page’ actually means ‘news that have “read more”‘).
Wow, seriously? Are we going to have to go through this for every page 2 story? Is it really that hard a concept to grasp?
Page 1 == editors write a blurb about the story, expanding it beyond a copy/paste summary. Note how these all have “read more” links.
Page 2 == “story” is just a copy/paste summary with a link to the real story on some other site.
Which essentially makes it problem with the editor. Why does OSnews have an editor who prefers to write editorials about everything and anything other than hard core technical OS stuff. I still enjoy OSnews but it seems to taking a far more ‘soft’ direction where shitty sitcoms get more editorial coverage than hard cutting edge technical developments.
http://i40.tinypic.com/2n7k1t.png
My comment was of a more systemic nature. As a rule the front page seems to get more of the fluff and far less of hard technical stories.
This story IS on the frontpage. That’s what I’m trying to say. You guys are getting knickers in twists over nothing but labels. We made a change recently that puts all page 2 items in the exact same place as front page items, so what is all this nonsense about?
That’s what I mean with the image.
But still very much a second class citizen. For what its worth I think the new page box in the middle of the page is more annoying than the old box on the side, but that’s just me.
If you want my opinion (and I’ll understand if you don’t) get rid of page 1 and page 2. Have all stories equal. If you wish to add “read more” to some and not others that’s totally cool, but don’t split the stories based purely on that.
I agree with previous poster too. The inline box sucks compared to the previous two tabs. Perhaps a “Most voted” tab for preferred stories wouldn’t do harm? Just another way to order stuff.
Good explanation. Sorry for the confusion.
Then is just plain sad that who maintains this place only writes articles about everything but operative systems.
Edited 2009-05-23 04:05 UTC
Thanks for getting this stuff up to date. The features for ZFS I’m mostly waiting for involve the ZFSBoot!
I can’t wait to get everything ZFS so my disk management gets really easy. Swapping out harddrives just becomes such a breeze.
Thanks for all the effort going into this.
btw, I think it sucks that out of 15 comments on this news, only 2 seemed relevant to the original post.
ZFSBoot is possible and available under FreeBSD. I would, however, recommend going the route of a gmirror-UFS boot partition to store rescue binaries on (besides the kernel of course). It could save a system from an unfortunate mistake