Today Jeff Roberson committed his patches to FreeBSD 9 for adding journaling to UFS. No more background fsck after unclean shutdowns! This is a major landmark in the history of UFS, with 11000 new lines of code (and about 2000 removed). Much of the work was done in collaboration with Kirk McKusick, the original author of FFS and Softupdates, under sponsorship form Yahoo!, Juniper and iXsystems. Jeff’s blog contains quite a lot of technical information of his work. There’s also information on the FreeBSD mailing lists.
I thought ZFS was FreeBSDs future?
I guess having something that won’t demand huge amounts of resources is good to have and it is cool to see that FreeBSD has catched up with Ext3 in the journaling department.
I like FreeBSDs RAID1 implementation, it does duplexing really well (not like Linux).
ZFS is and at the same time isn’t the FreeBSD future. It isn’t because of the license. ZFS is licensed under the CDDL. The BSD people have been rewritting GNU userspace tools only because of the license, so having a non-BSD filesystem (the heart of a unix OS) is certainly something they don’t like a lot. Also, while ZFS is cool, it can make FBSD users wonder – “hey, and why not use opensolaris instead”
But, at the same time, ZFS is the future of FreeBSD, because it’s the only filesystem that keeps FreeBSD as a good server OS in many workloads.
IMO FreeBSD should consider seriously a substitute for UFS (maybe hammer?).
Edited 2010-04-25 20:51 UTC
That’s right, BSD people don’t like messy licences [in an OS base], but that applies specifically to OpenBSD. FreeBSD and NetBSD tend to care less for licensing.
I don’t think that *BSD FS needs to be replaced by anything. It’s one of the most rock-solid [and proven by many years] FS I’ve ever used and I used/am using quite a lot of FSs [not a single data loss on UFS/FFS. ext2/3, NTFS made it couple of times].
Edited 2010-04-25 23:28 UTC
It is solid. But the one you admire, OpenBSD, doesn’t even support journaling….
You know, when it comes to OpenBSD, it doesn’t matter how many unproven ‘features’ OS has. It’s all about stability and security here
And yes – I am using OpenBSD among other OSs and it performs very well on its default FS.
Needless to say I don’t need journaling at all, mostly because of its nature – it simply makes my data more vulnerable to loss and – as some of other posters have said – it doesn’t really make that much difference [ca. 1 min? not much, honestly].
“I have done some fsck benchmarking. I recovered an 80% full 250gb volume that was doing a buildworld with 8 parallel processes in .9 seconds. The traditional fsck took 24 minutes.”
http://jeffr-tech.livejournal.com/
ever fsck couple of TB?
Edited 2010-04-26 12:53 UTC
explain how journaling makes data more vulnrable to loss.
I was describing my own ‘adventures’ with ext3 mounted in compat mode as ext2. It destroyed my data.
Let’s hope UFS journaling will work a whole lot better.
so… a compatability mode of a file system destroyed your data, not journaling.
No I think their issue is with how included GPL software restricts the usage of FreeBSD. GPL 3 has requirements for hardware companies that FreeBSD developers have never supported.
Not anymore with Oracle at the helm. The future of OpenSolaris is uncertain and at most Oracle will provide minimal funding for p.r. reasons. OpenSolaris could get the axe at any moment which makes FreeBSD a much safer choice. Any fork of OpenSolaris would likely stagnate. OpenSolaris will attract few developers when its best features have already been ported to FreeBSD.
“The future of OpenSolaris is uncertain and at most Oracle will provide minimal funding for p.r. reasons. OpenSolaris could get the axe at any moment”
That is not true. Oracle has officially stated it’s continued support for OpenSolaris. In the future, OpenSolaris will be like Fedora, and Solaris will be like RedHat. Both will be developed, and none will be axed. Oracle/Sun competitors are spreading the FUD extensively, please stop it because it is not true.
http://itmanagement.earthweb.com/osrc/article.php/3867771/OpenSolar…
Let’s face it, the most innovative operating system that everyone wants to copy is Solaris. In the future, the development pace will not stop. It will continue. In my opinion, Solaris is most innovative today.
Oh FUD huh? So does this look like a supportive move?
http://www.katonda.com/blog/1058/oracle-puts-another-nail-open-sola…
So because Oracle is not offering OpenSolaris CDs for free shipment anymore, it means Oracle is killing OpenSolaris? Are you serious? Jesus.
You can download OpenSolaris CD for free. So Oracle made other decisions than Sun, how to distribute OpenSolaris. That means nothing. Oracle has officially and publicly stated that OpenSolaris will continue. You know that, and if you still tells everyone that OpenSolaris will be killed – then it is FUD.
I heard that RedHat has changed fonts in their manual to a smaller one – that must mean that RedHat is going to be killed. I know that RedHat is officially supporting Linux, but hey, what they say offically means nothing, right? It must mean that RedHat is probably killing Linux and migrating to OpenSolaris.
Oracle says something, you write that Oracle is telling the opposite. That is pure FUD.
I didn’t claim it was proof that they are killing OpenSolaris. The author wrote that. But it is a very bad sign.
Oracle has 17.5 billion in cash in the bank. Yes in the bank.
How much was the OpenSolaris CD program costing them? I doubt it costs more than a thousand dollars a year. The fact that they are cutting pennies here speaks volumes. OpenSolaris is not an important project to them.
My God you really have no idea as to who bought Sun. These are not pony-tailed hippies that will continue funding pet projects just because they’re open source. OpenOffice at least makes sense to fund as a long play against MS but OpenSolaris is really looking like an unwanted stepchild.
Larry is wise enough to not outright kill a project like OpenSolaris. The tech press would portray him as anti-open source. If anything he’ll slowly defund it by diverting resources to Solaris.
OpenSolaris is going nowhere and at best can hope to be a testing ground for new Solaris technologies. FreeBSD has never looked better in comparison and anyone who invests in OpenSolaris at this point deserves their fate.
Oh and here’s some more of that wonderful support the OpenSolaris team is getting:
http://www.businessweek.com/idg/2010-04-15/opensolaris-leaders-unne…
Bad sign for what? Are you serious? Oracle changed distribution channel for OpenSolaris, so what? Oracle changed logo from Sun to Oracle, isnt that also a very bad sign? I heard that Larry had a hair cut from a mediocre saloon, isnt that a very bad sign? Because Larry is very rich, if he doesnt use the most expensive saloon, it must mean the expensive saloon is going bankrupt, yes?
Lots of authors are pro-Linux and they loathe Solaris. They take every chance to spread FUD about Solaris. I count you to that camp. You dont even have nothing of substance to base your FUD on. Sun did things the Sun way, Oracle does things the Oracle way. Now Oracle is changing Sun to the Oracle way, and that change is a sign of OpenSolaris is slowly getting killed? FUD, or what?
Just because Sun distributed OpenSolaris CDs, it doesnt mean that Oracle must do it the same way. Maybe Oracle never distributed CDs, ever? Then what? If Oracle uses SAP for business, and Sun used some other company, is it a bad sign that Oracle only wants SAP? Jesus.
Do you have any proof or is just your wish?
Jesus. OpenSolaris IS a testing ground for new Solaris technology. That is the outspoken PURPOSE of OpenSolaris and has always been. I wrote that OpenSolaris is similar to Fedora, and Solaris is similar to RedHat. Good that you finally understand the purpose of OpenSolaris at last, though.
After the bugs in new functionality have been ironed out, OpenSolaris functionality gets backported to Solaris. For instance, ZFS arrived first to OpenSolaris, then it got backported to Solaris.
It would be very dumb to kill OpenSolaris, and use Solaris both as a test ground and in production. First you need to iron out bugs. You never do that in production. To spread word that OpenSolaris will be killed in a couple of years is just: FUD. But it is true that OpenSolaris is a testing ground, please say that, instead of pure FUD.
Good for FreeBSD and congrats to the FreeBSD team. FreeBSD is a very good OS, yes. Much better than Linux. If there were no OpenSolaris, I would have used FreeBSD, without a doubt. In the future more of Solaris technology will be ported to FreeBSD, making it even better. The brilliant Solaris engineers are not resting.
So what? Sun had a roadmap and spoke publicly about it, Oracle is silent until they release (just like Apple). It is just different policy.
Look, Oracle does things differently, it is not a sign of OpenSolaris is getting killed. Especially when Oracle officially supports OpenSolaris. When you gladly jump on every anti-Solaris article you are just helping the FUD.
Maybe a strong OpenSolaris is in your interest too. How else would FreeBSD get new cool innovative technology? From the Linux camp? They are just copy cats, and have produced nothing innovative. BTRFS is a ZFS wannabe. DTrace also has Linux copies. etc. Sun contributed more open source than any other company. Without Suns contributions, the world would look differently.
http://www.oracle.com/webapps/dialogue/dlgpage.jsp?p_ext=Y&p_dlg_id…
🙂 So Oracle does ship free CDs for linux.
Thanx for posting this. That weakens my argument, yes. But I still dont think it is optimal to draw conclusions from if a alfa version is not distributed on CD anymore. Yes, OpenSolaris is not in Beta version, it is in Alfa version – and OpenSolaris is not sold. You shouldnt draw any conclusions from that, I think. It is not a sold product yet.
Oracle Unbreakable Linux is sold with full support. It is a product that is sold. OpenSolaris is not. OpenSolaris has no support, unless you run it on Sun hardware.
I’m giving you my opinion which is that the future of OpenSolaris is uncertain and doubtful. That’s entirely different from spreading FUD as a way of trying to deter interest. I’m certainly not pro-Linux or anti-Solaris. It’s just my opinion, deal with it.
I was talking more about it becoming an eternal beta and not something that Oracle wants anyone to run in enterprise.
There’s a downside to that which is that a buggy OpenSolaris could give Solaris a bad name. If OpenSolaris is just as good as Solaris it will cut into their own sales.
Officially supporting a product can mean keeping it alive for p.r. reasons while directing resources elsewhere.
I have no problems with your opinions. Everyone is entitled to have an opinion. But when you state something false as a fact, then I have a problem. Facts can be wrong, opinions can never be wrong. When you write that OpenSolaris will be killed any moment then it doesnt sound as your opinion or wish. Please be clearer in the future with what is your subjective personal opinions.
Jesus. But Oracle (nor Sun) does not want anyone to run OpenSolaris in Enterprise. That role belongs to Solaris!
I really dont understand your objections. It is like I say “Fedora is getting axed soon, because it is just a test ground, and the company RedHat doesnt want anyone to run Fedora in Enterprise – therefore Fedora is getting axed soon. Anyone who invests in Fedora deserves to die”. But that is the PURPOSE of Fedora – to be test ground, and not run in Enterprise. Jesus.
The problem is not that alfa version OpenSolaris is giving Solaris a bad name (Solaris reputation is far to good for that) – the problem is more like people like you, who dont have a clue about OpenSolaris/Solaris – spreading FUD which is incorrect and not true.
There are companies that run OpenSolaris in production today. OpenSolaris is now in alfa version. Try to run Haiku or any other alfa version OS in production. The fact that you can run an alfa version is a testament to good stability. No worries about bad name. And besides, let Oracle worry about that. I appreciate that you worry about that, though.
As I said, OpenSolaris is the testing ground for the Enterprise Solaris OS. If Oracle killed OpenSolaris, then there would be no Solaris 11. OpenSolaris is used for developing next gen Solaris 11.
Do you finally understand that Oracle will not kill OpenSolaris, because Oracle expects to deliver Solaris 11 in the future? Now, can you stop spread FUD about OpenSolaris is getting killed soon? If that is your wish and hope, then say so. But dont try to make it look like facts, by supplying weird links (that show that there are no more CDs of an OS in alfa version). If you look at wikipedia article on FUD, you will see that you are exactly doing that: FUDing.
“FUD is generally a strategic attempt to influence public perception by disseminating negative and dubious/false information designed to undermine the credibility of their beliefs”
Oracle has said they are supporting OpenSolaris, next gen Solaris. Still you are telling everyone it is getting killed soon.
Yeah right. You write:
“…anyone who invests in OpenSolaris at this point deserves their fate.”
“OpenSolaris could get the axe at any moment which makes FreeBSD a much safer choice.”
I understand that you are worried that FreeBSD people will migrate to OpenSolaris. I understand your worry. But hey, FreeBSD has been around for many years, and it is much better than Linux. I think you dont have to worry. And besides, FreeBSD benefits from OpenSolaris. In the future there will be more new cool OpenSolaris features that will be ported to FreeBSD. OpenSolaris developers will not rest, there will be more new functions. Sun has always been innovative.
If you posted that on any OpenSolaris message board, they’d rip into you. OpenSolaris is not meant to be alpha. It’s meant to be a production open source operating system. No different than Linux, FreeBSD, OpenBSD, or the others.
I agree, the CD thing is not a big deal. However, Solaris doesn’t need OpenSolaris to exist or test. It didn’t need it before, and it doesn’t need it now. It’s got many purposes, including as a public playground for some features that may or may not end up in Solaris, but they could go back to private. Another purpose of OpenSolaris is to an ambassador for Solaris, to slow it’s downward trend. Solaris was hemorraging web servers and app server installs, and was losing on big-iron as well.
But could Oracle kill OpenSolaris? Yes and no. Sun released the code under CDDL, so there’s no way to put the geenie back in the bottle. However, if Oracle stopped funding the organization and prohibited Oracle employees from contributing to it, it’s hard to imagine that OpenSolaris wouldn’t be mortally wounded in terms of momentum.
I don’t know stats, but it doesn’t occur to me that OpenSolaris adoption is increasing that much. At best, it seems stagnant. It never really caught on the way that Sun had hoped (if it had, Sun probably wouldn’t have had to sell out to Oracle). Anything (such as Oracle withdrawing resources) that Oracle does could cause its anemic adoption rate to go into steep decline.
All you have to do is look at Oracle and MySQL. While Oracle hasn’t killed MySQL, they certainly haven’t done it many favors.
Plain wrong. Common agreement is that OpenSolaris is buggy right now and not meant for production – in it’s current state. Of course the forum people expect the bugs to be ironed out in the future. But as of now, OpenSolaris doesnt cut it for production. This is general agreement. Everyone knows this.
Sure, but now Sun did a change. Instead of only letting some Enterprise customers to try out beta versions of Solaris, they have opened it up. No big deal, actually. But Sun gets more feedback from the customers regarding the beta and alfa versions.
Anyway, OpenSolaris IS buggy as of today (everyone knows this) and it is a play ground for next gen Solaris.
Edited 2010-04-28 08:19 UTC
That is far from the common agreement, especially if you ask the OpenSolaris developers. It’s only your opinion that it’s not ready for production, but that’s not the widely held opinion. It’s got bugs, but so does Solaris, Linux, and all other operating systems. They get identified, and they get fixed.
They get more feedback, but they could do without it. Hence, they could kill OpenSolaris and not care all that much. Solaris in no way needs OpenSolaris to succeed, thrive, or even exist.
Not true. Some people consider that as a matter of opinion, but more consider it a production operating system. The stated goal of OpenSolaris is not “buggy, alpha release”, but instead a solid operating system to compete with the likes of Linux, Windows, and the BSDs.
“That is far from the common agreement, especially if you ask the OpenSolaris developers. It’s only your opinion that it’s not ready for production, but that’s not the widely held opinion. It’s got bugs, but so does Solaris, Linux, and all other operating systems.”
No, this IS the common agreement. OpenSolaris is not for production, it IS buggy. It IS a test ground. New functionality gets introduced every second week. You can impossibly have an Enterprise company based on something that changes so rapidly. You need to test every release, very well. OpenSolaris releases every second week, is hardly tested at all.
Seriously, I dont get it. How can you believe that Sun claims that an alfa OS released every two weeks, without testing, is for production? It is not. This IS the common agreement. Ask the mail list if you wish.
Sun/Oracle charges for Solaris. OpenSolaris are for the masses. Solaris are for Enterprise.
You’re mistaken on the nature of OpenSolaris releases versus builds. They do weekly builds (even nightly I think), just like a lot of other vendors. That’s not the release. OpenSolaris has a distribution, 2009.06 (project Indiana) with updates you can download for it just like Ubuntu has for 10.04, or FreeBSD 8.0.
Of course the nightly builds are not meant for production. But the 2009.06 builds are.
From the OpenSolaris page: OpenSolaris Operating System: This is a community-developed binary distribution of an operating system based on the OpenSolaris source code. Development efforts take place at Project Indiana on opensolaris.org. The distribution runs on SPARC, Intel, and AMD processors on the server and desktop and as a storage platform. It is free to use, modify, and redistribute. Support is available from Sun Microsystems.
Whatever. If you think that buggy OpenSolaris released every 6 months is for production, you are free to think so. But I know, that it is not for production. Solaris 10 is for production. Not OpenSolaris. I suggest you ask on the forums. Ask something like “is OpenSolaris for production, or Solaris 10?”. Solaris 10 is for production. OpenSolaris is a play ground.
What gives you the impression that OpenSolaris is buggy? Or at least, buggier than other operating systems?
Few people think OpenSolaris isn’t production ready. Those that do, don’t seem to have have anything to back it up or than pedantic nitpicks or non-technical biases. Yours so far is “it’s buggy”, as well as other claims that were disproven (such as the weekly release issue).
Ask Ben Rockwood (Cuddletech.com), or this article from two years ago: http://www.linux.com/archive/feature/147358
or this: http://www.sunhelp.org/pipermail/geeks/2009-March/066475.html
It’s not buggy (or rather, it’s the same relative bugginess to other operating systems. Yes, even Solaris has bugs). OpenSolaris has been around for 5 years and it’s a solid operating system. Solaris and OpenSolaris are meant for different uses, for different customers. Solaris has a solid reputation, and a better fit for the Enterprise and big-iron market. OpenSolaris is probably a better fit for ISVs, storage, cloud, appliance, web, and app serving.
The fact that it’s a solid operating system doesn’t mean it usurps other operating system, or is a better fit for what you’re doing. But “buggy” is not a valid or truthful reason not to use OpenSolaris, especially when “i just don’t like it” is a perfectly valid reason not to use an OS.
Edited 2010-04-29 22:04 UTC
Uhm, what? BSD projects don’t like the GPL. There’s nothing wrong with the CDDL, which is why a lot of OSol tech has made it’s way into FreeBSD. There’s no “everything must be BSD-licensed” mantra in the BSD projects. There’s just an unspoken “let’s avoid the GPL” motto.
Why use something completely foreign when one can use all the management tools and frameworks (like GEOM) that you are familiar with?
I’d have thought the reverse was true. If FreeBSD didn’t have ZFS then some FreeBSD users might consider OpenSolaris.
FreeBSD is a very different beast to OpenSolaris – so I couldn’t see FreeBSD users (ie users that would otherwise be happy with FBSD) migrating to OpenSolaris unless it was for some features they felt they needed but FreeBSD didn’t offer
I know everyones computer requirements differ, but for me ZFS would be a major selling point in my returning to FreeBSD. Unfortunately I’ve not felt persuaded about the stability of FBSDs current implementation
ZFS is the future of FreeBSD amd64 systems with at least 2gb RAM, preferably 4. However, there are still a fair amount of i386 systems out there or amd64 systems with less than 2gb RAM. For these users, the only option is UFS and journaling softupdates are great news for these people, considering how fragile UFS can be under certain circumstances.
Now I wish someone did extensive performance and reliability comparisons between gjournal and the new journaled softupdates. The big advantage of the latter is the ease with with they can be turned on/off on an existing running system.
I hope not. ZFS has too many problems, we used it in production for many years.
It just sucks …
I modded you down as a troll because without providing any examples or links that’s what you are.
You want proof ?
ok bastard (I call you bastard because you moderate me wihtout knowing ZFS)…
first, try to use it with high level of I/O and not you the computer under your bed …
Now:
– you CAN’T remove a disk. No problem you say ? I say no. Try to migrate your LUN from one bay to another and you are doomed. Try to consolidate your LUNs with less and bigger one ? You can’t. And don’t tell me it’s coming, they are saying that since 5 years now …
– copy-on-write fragment the filesystem over the time and guess what ? You can’t do anything about it. No tool (and prefetch don’t do it for high I/O).
– ZFS just kill your san: our storage team told us it was eating 30% of the total I/O only for one host.
– Mirroring with ZFS is crap. ZFS only knows that a mirror is corrupted when it tries to access the data and it will only correct the corrupted data he tries to access. (great no ? what just happens if you lose the good one ?).
The only way to detect that the disk is corrupted if you don’t access this data is to use scrub, which kill you disks / san (cf over). And yes it happens to lose a good disk and then it came back (ie you lose a path on a san). Now to resync your disk you have remove it and take it back, but it has to resync everything from scratch…
And I don’t speak about ZFS cache problems we had, and the problems we got with those crappy solaris zones…
The fact is that ZFS eats a whole bunch of memory. Don’t know about disk I/O, this might be true in some cases.
I really don’t think your comment should ever be modded down just because you added “ZFS sucks” in your comment. It’s obvious, that it sucks to YOU. Maybe you shouldn’t have use a stron language, or maybe the guy who modded you down should reconsider his participation in public discussions.
Regards to all
Well, you usually use this langage when you spent night and day on sun craps and that somebody tell you that it’s not true …
“first, try to use it with high level of I/O and not you the computer under your bed …”
You have to configure ZFS differently than other file systems. ZFS is not perfect, but you need to know what you are doing.
It is well known that all raid solutions have trouble achieving high IOPS, but it is possible to circumvent that problem with ZFS. One way is to use SSD as a ZFS cache. Then you can reach good performance, for instance CIFS ~3GB/sec over a 40GBit NIC and 300.000 IOPS. This ZFS server is basically a bunch of SATA 7200 and SSD and quad core CPUs running OpenSolaris + ZFS:
http://blogs.sun.com/brendan/entry/iscsi_before_and_after
Sure, you can not remove a disc. It may be a problem. But many Enterprise companies are mostly interested in adding storage capacity.
Copy-On-Write fragments discs, yes. But ZFS takes measures to minimize it. It collects data until 7/8 of RAM is full, or 30 seconds have passed – before writing all data in one go. That minimizes fragmentation. Hence, lots of RAM is preferable for a ZFS file server. ZFS is mostly targeted for Enterprise companies. If you have less RAM and less needs, then other file systems may be better for your needs.
The only way to detect that a disc is corrupted is when ZFS tries to access the data – how is that bad? How can ZFS work differently? How can a filesystem know that data is corrupted, without even trying to access the data? Of course you need to access the data first, to see that it is corrupted! Otherwise it would be magic. This applies both for ZFS mirrors or ZFS raids. You need to access the data first, to see that it is corrupted. If you think ZFS mirrors sucks, then also ZFS raid sucks – because ZFS raid also requires you to first access the data to see if it is corrupted.
I get the impression that you are not handling ZFS correctly. I mean, you talk about small issues, but dont even mention ZFS biggest advantage, which no other normal file systems offer. That is the ONLY single reason to use ZFS. I am quite sure you dont even know what advantage I am talking about…
dude. Are you serious ?
You are taking a Sun blog to back up that ZFS is cool.
You are joking right ?
What I learnt from Sun since Solaris 10 is not to believe their words…
No, I am showing the Sun blog to show that ZFS can reach extreme performance and IO. You claim that is impossible with ZFS:
I am only disproving your claim with my link. Or, do you doubt the performance numbers quoted in that blog? They are lies?
This is actually interesting. Could you explain more on this? Sun has been lying to you? In what way?
This is also interesting. You claim that Sun hasnt been able to solve your problems? Is it true? What was your problems?
[quote]I get the impression that you are not handling ZFS correctly. I mean, you talk about small issues, but dont even mention ZFS biggest advantage, which no other normal file systems offer. That is the ONLY single reason to use ZFS. I am quite sure you dont even know what advantage I am talking about…[/quote]
Yeah, it must be that. And I believe that Sun support experts are not handling ZFS correctly too because they never solved our problems …
Very few RAID controllers (I haven’t personally seen one that does) allow you to remove disks from an array. Why should software RAID be any different?
Not yet. But there is work underway to fix this.
And … you can prove 100% that it is ZFS doing the I/O and not the applications running on top?
A hardware RAID controller doesn’t know data is bad until it reads the data, or runs an auto-verify in the background. Just like ZFS, which can run a scrub in the background. In fact, nothing in the world can tell a block of data is corrupt without first trying to read it. At least with ZFS, the corrupted data can be detected and re-written automatically. A hardware RAID array just notes that the sector is bad.
And the only way for a hardware RAID controller to detect that a disk is corrupt is to run an auto-verify in the backgroun, which either kills your I/O or takes forever, depending on the settings on the controller.
No, it only re-silvers the data. ZFS doesn’t re-silver empty space. In comparison to hardware RAID controllers that have to sync every block of the new drive, regardless of whether it’s in use or not (the controller doesn’t know the details of the data on disk).
Can’t comment on Solaris Zones. Never used Solaris.
However, it sounds to me like you don’t really understand how ZFS works, and have been trying to use it like a hardware RAID setup. Which is just wrong.
You know, not everyone uses ZFS. IMHO it’s more suitable for the servers. As the legacy of sun comes the heavy memory footprint, which isn’t a good thing for the desktops. Besides: this contribution will be – probobly – adopted to other BSDs as well [although I don’t think it’s a good idea, as journaling makes your data more vulnerable].
If I understand the blog posts(“Come Play, My Lord.”) this is not actually storing any data on the journal but recording a few of the corner cases not covered by softupdates.
The real question is if this is needed. You are sacrificing overall performance for very little gain. You only have boot once after a forceful shutdown. A minute more or less won’t make a difference IMO.
Matt Dillon has stated he doesn’t want to update UFS in DFlyBSD, as things are nice and stable right now. Plus, UFS is really only used for /boot, with everything else being HAMMER.
I was actually thinking about NetBSD/OpenBSD. DFBSD stays unnoticed for most of the time [which is good IMHO. No hypes, just a solid work of DFBSD team].
I guess someone could say about time but I think this really is great news. I suppose FreeBSD is going for incremental updates rather than any radical changes.
9,000 lines of code (although as Jeff says this number may be misleading if seen by itself only) is too much for a single feature, IMHO.
Kudos nonetheless.
It’s a complex feature. 🙂
Anyway, consider that FreeBSD’s source code is very tidy and well documented, so 9 kLOC don’t seem that much if you take into mind that things like empty lines (for visual separation) or comment blocks (for explaination) also add to LOC.
Especially when you consider the coding style, as outlined in style(9), where { and } are only separate lines by themselves, and if-statements are always multi-line.
LoC is really a useless, and pointless metric.
It is useless when seen alone and not contrasted with the LOC count of another filesystem.
For example this is very illustrative IMHO:
http://blogs.sun.com/eschrock/entry/ufs_svm_vs_zfs_code
Or, compare freebsd softdeps (11K LOC) with freebsd gjournal (<5K LOC), or freebsd softdeps with netbsd wapbl (5K LOC).
And yes softdeps aren’t the same as journaling, but they serve the same purpose
Anyway, as I said kudos to the developer — we will all be enjoying his contribution!
Cheers
I cant understand why FreeBSD doesnt use Hammer instead of USF or ZFS? Hammer is BSD licens and more suitable for desktop use.
Hammer is also relatively new, and therefore has has much less testing than more mature filesystems for starters.
Furthermore Hammer as it is in DragonFly relies on a ton of changes made over years to the VFS layer alone and likely changes to many other kernel subsystems that have not occurred in FreeBSD as they were not needed by it and would thus need to be ported before any serious attempt to get Hammer working could get underway.
And before you mention the fuse ports of Hammer that exist for Linux and Mac OS X, realize that they are to allow read-only access to Hammer filesystems and are a few versions behind at any rate.
Hammer is in version 4 already and has been heavily commited ever since day 1.
ZFS is still one of the topics that causes the bulk of emails sent to @stable and @questions …
Are you sure about that?
Try asking in hammer-request at lists.dragonflybsd.org .. People interested in porting HAMMER to other operating systems should contact Matthew Dillon at dillon at backplane.com
Even if it is opensource, Linux still has problems working with UFS .. I wouldn’t count on Linux being the one to take Hammerfs outside of DragonFlyBSD’s world..
Edited 2010-04-26 07:06 UTC
Maybe because they are strongly reluctant to import the coding marvel of the very same guy they had kicked out?
Maybe because the code doesn’t have a “FreeBSD feel to it” .. like in Murenin’s port of the OpenBSD sensors framework to FreeBSD (GSoC2007/cnst-sensors)?
Maybe because they decided that with all the money, devel time and debugging effort they spent porting ZFS, ZFS should be the FS you should use regardless of wether it is or it is not a better FS than Hammer?
Maybe because of the same reasons why the keep Sendmail in base instead of moving it to ports and import DMA or OpenSMTPD to base (which are _nothing_ but historical reasons)?
Go ask FreeBSD satus quo … they probably have a rock solid and scientific explanations about it like “Because it has been there from the beginning” or something lines along that way ..
Anyways, my congrats and sincere thanks to Jeff for his work. It will be highly aprecciated in my ZFS free world
Edited 2010-04-26 07:04 UTC
That is a good reason!
I didnt think of that, this early in the morning here in sweden. I was just thinking BSD = BSD.
HAMMER isn’t in FreeBSD because nobody had ported the code to FreeBSD. In addition, there’s concerns about scalability and reliability of the code, and how well it will fit into FreeBSD’s model for multitasking. It is a big project, and it isn’t clear there’d be a big win from completing it. there was also a lot of hype when it came out that turned out to be overblown when people took a first look at it, which has lead to a reluctance to give it another look.
So until somebody shows up with a port of HAMMER, it is silly to talk about “why” it isn’t in FreeBSD. FreeBSD has evolved significantly in one direction, to support better MP scaling, and DFBSD has evolved in other directions. This means the porting effort would be large. Much like porting between Linux 2.0 and 2.6 in many respects, to draw an analogy.
Warner
All the BSDs suck and are 1,000 times behind Linux.
Long live Linux!