What is Digg?76 Comments
- pinetree, on 10/12/2007, -0/+41"Reiser File System is the default file system in SUSE Linux" -- not anymore. They switched the default to ext3 in OpenSUSE 10.2. Also, there should probably be separate entries for Reiser3 and Reiser4.
- nicku, on 10/12/2007, -0/+37Yes, he has been charged with her murder:
http://en.wikipedia.org/wiki/Hans_Reiser
SuSE switched to Ext3 two days after he was charged. - justinjacobs, on 10/12/2007, -0/+30Yep
http://en.wikipedia.org/wiki/Everything - azbmr, on 10/12/2007, -1/+27I heard a rumor that Reiser killed his wife (or at least is under suspicion of doing so), which would make ReiserFS' devel grind to a halt. Might be best to stick with ext3 or XFS.
- archlich, on 10/12/2007, -0/+22http://en.wikipedia.org/wiki/File_system_comparison
^ the real file system differences - swanny89, on 10/12/2007, -2/+20thats an unfair statement. I just installed kubuntu, and as a brand-new linux user I've found that everything is either extremely easy (updates, printers), or very hard (video drivers, any application not supported by apt-get).
- NewChar, on 10/12/2007, -3/+21That's inaccurate, *insert Linux success story here*
- stalefries, on 10/12/2007, -2/+18But is that a dumb joke?
- inactive, on 10/12/2007, -0/+14The NWFS is awkward to use in public. : /
- geronimo, on 10/12/2007, -0/+13The license ZFS is under is not compatible with the GPL v2, and that was by design. Sun doesn't want ZFS working under linux, from a business standpoint it wouldn't be good. If they move it to GPL v3 like they are hinting at then it couldn't be in OS X.
- GMorgan, on 10/12/2007, -2/+14Ext3 is more stable anyway. Even Gentoo warns against using Reiser3 and totally refuse to even consider Reiser4 (when you pull in the kernel there's a message saying how utterly stupid Reiser4 is and if you must be insane R3 is at least tolerable).
Anyway I'll stick with the tried and tested for now. Yes file systems are one of the most important aspects of Linux given that everything is a file but Ext3 is good enough. - muffinmanpoo, on 10/12/2007, -0/+12There's a Wikipedia article for everything, isn't there?
- Rammsteined, on 10/12/2007, -0/+12Does POSIX compliance mean anything to you?
- tehbored, on 10/12/2007, -3/+14Linux isn't is plenty simple! I bet you've never even tried it.
- burke, on 10/12/2007, -2/+12Too bad Hans Reiser had to go and kill his wife. Apart from that really sucking for his wife, it's unfortunate for Linux users, as reiserfs was just starting to gain ground on (the slightly inferior in most respects, other than stability at the moment, as mentioned above) ext3 as the predominant default file system.
- aRgusChung, on 10/12/2007, -2/+11yeah, "THE SAME SO IT DON'T MATTER THANK YOU LINUX IZ TEH PWN#@#" until you need to do disaster recovery and need to pull data off of your drive from another OS.
I've had RFS machines ***** themselves and had a bitch of a time doing drive recovery from OSX. EXT2/3 is well supported on pretty much every platform. Even "Microsoft LOL".
(and yes I know Yarg or whatever is around for windows, but I shouldn't have to have a windows machine around to support Linux) - GMorgan, on 10/12/2007, -0/+9Don't get me wrong Reiser3 had a lot of potential and Reiser4 even more so but the crucial factor is always how likely my data is to still be there tomorrow. With some effort I could foresee a Reiser3 that had at least equal space next to Ext3 4/5 years from now but it does need that level of battle testing.
There's nothing to say someone else couldn't take Reiser on. I'd really like to see R4 get finished and have the code be brought upto kernel standards for inclusion. It just isn't there yet. - inactive, on 10/12/2007, -4/+13Those are not hair questions.
- deadbaby, on 10/12/2007, -0/+8I can't really recommend any casual Linux user stray away from EXT3 these days. There was a time when EXT2 was the standard and it really sucked so I tried out ReiserFS and it eventually ate several of my file systems. Since then I've always stuck with EXT3. Not only is it the most tested file system out there but it has the best data recovery tools. I've had a few incidents but they were easily fixed by running a few commands. ZFS (or a 100% OSS clone of it) may be something to look at down the road but until then EXT3 is the best bet.
- muramasa, on 10/12/2007, -0/+7I'm surprised that they (or someone else) didn't fork ReiserFS. Even if Hans is found not guilty his name has been dragged through the mud, and there will always been suspicion. It had a good future and continued development would be good for Linux in general.
- tarpit, on 10/12/2007, -0/+7You can get read-write access with driver NTFS 3-G
http://www.ntfs-3g.com
The NTFS-3G driver is an open source, freely available NTFS driver for Linux with read and write support. It provides safe and fast handling of the Windows XP, Windows Server 2003, Windows 2000 and Windows Vista file systems. Most POSIX file system operations are supported, with the exception of full file ownership and access right support. - maninblac1, on 10/12/2007, -1/+7AFS is a nice idea in theory, but it is attrocious in practice. It's so bad as a system for our clients to deal with we're moving away from AFS and onto believe it or not, DFS this fall.
- Ibanezfoo, on 10/12/2007, -0/+6We use XFS on our file servers.... nice and fast. Seems stable for the last 3 or 4 years that we've used it.
- tehbored, on 10/12/2007, -0/+5I remember when I tried to get the ntfs filesystem on my external hard drive to work properly on linux. Then I remembered that there was no point in trying since fat32 is WAY easier because you don't have to do ANYTHING to make it work.
- sotopheavy, on 10/12/2007, -0/+5I really wish this information was included into the partition process when installing linux. Hint Hint.
I also wish you could have the option to go through typical installations (as a slide show on the live cd) during the installation process.
If you're experienced you can skip it, if you're new it would be a ScienceSend! - GMorgan, on 10/12/2007, -0/+5Pity that driver depends upon recursion otherwise we'd see someone try to move it into the kernel by now* with all the speed and stability gains that would entail. Still at least we have a working NTFS driver and in time someone is certain to refactor the code to depend upon non recursive control structures.
*the algorithms I mean, the actual code base can't be merged in as easily as all that. - duhblow7, on 10/12/2007, -1/+6GMorgan:
Do you have a link re: Gentoo advising against reiserFS (3)?
I've ONLY used reiserFS 3 as my file system for all my gentoo boxes, although I do use ext3 for the /boot on them all as well. - gharding, on 10/12/2007, -4/+8Just wait for a Linux port of Solaris' ZFS. Goodbye ext3!
- inactive, on 10/12/2007, -3/+7
I prefer the Reiser file system because it has killer performance and it was also developed by a Killer. - dtfinch, on 10/12/2007, -0/+3I spent an inexcusable amount of time browsing filesystem reviews where I last worked, when we decided our new fileservers would run Samba on Linux instead of Windows.
I really like JFS, but when it really matters I put of my trust in ext3, simply because it gets the most developer attention and the most support. Both seem to fair pretty well in a power outage. Ext3 beats JFS at full directory traversal (the find command takes less time), but JFS wins at most other benchmarks.
I don't feel like I can trust XFS, because I've seen a number of reports of it not fairing so well in a crash/outage. https://launchpad.net/ubuntu/+bug/37435 for a somewhat recent example. And the code size worries me, being about 3-4x larger than most of the other journaled filesystems available on Linux.
I have no feelings about Reiser3. It's a good all around performer, but I haven't used it much except with slackware, where it's the default.
I worry about Reiser4's cpu usage in published benchmarks. Most of the time, it wouldn't matter, and Reiser4 would be the fastest, but it seems like something that can easily grow to be a bottleneck. I expect that Reiser4 will be a source of inspiration for new filesystems for years to come, even though its own future looks bleak. - majinboy, on 10/12/2007, -0/+3LVM is a method by which multiple partitions can be seen as one partition. This article talks about the underlying filesystem a partition may use.
- geronimo, on 10/12/2007, -2/+4There's also AFS(Andrew File System) and Coda, both open source.
AFS is for large scale filesystems, Coda is like AFS but it allows for detached storage. AFS has allowed for fancy directory-based ACLs since the 80s. - mpeters13, on 10/12/2007, -1/+3Article was well laid out and gave the right amount of information about each format type w/o biases. More articles please!
- int19h, on 10/12/2007, -0/+2"Not Workable for Suits"; an alternative to "Not Suitable For Work"? ;)
- swordedge, on 10/12/2007, -0/+2Remember what JFS was developed for. The AS/400. They needed a robust file system with good speed in an environment that often had thousands of users on it. When it was ported to OS/2, they still booted with HPFS but used JFS for data storage. JFS is great for the main storage drive for that but proved slower then HPFS for OS type operations.
The more powerful tool is probably LVM. I don't know if Linux can do this but on OS/2, you could add a new drive to your system, partition it, and add it to an existing JFS volume.... ON THE FLY. That is no need to shut anything down. Mind you, in a proper business environment, I'd wait till like two am to do that but still. This assumes you have a hot pluggable drive system. - KIERANMULLEN, on 10/12/2007, -1/+3Why isnt LVM talked about?
KieranMullen - DrDabbles, on 10/12/2007, -0/+2WTF? I keep reading these people posting "Won't reiser development stop now?" Why? Because he may or may not go to jail? Hi, Welcome To Last Decade. This is OPEN SOURCE. That _basically_ means any one of us (though, probably not the people making those comments) could pick up the sources and do whatever we want with them. INCLUDING but not limited to continued development / support.
Most of the problems with ReiserFS 3/4 were Hans Reiser. He basically out-and-out called linux broken because it uses 4k instead of 8k stacks. Reiser4 _requires_ 8k stacks. VFS is built to use 4k stacks. So, any kernel patched to use Reiser4 is patched to use 8k stacks. There are arguments going both ways, but Hans was too self-absorbed and full of himself to see anybody else's side of the argument. There are _numerous_ threads between Hans and the main developers on the Kernel Mailing Lists going back and forth about why each side is wrong. To be frank, Hans didn't like the feeling of being ganged up on, and the main kernel developers simply thought changing the entire stack allocation scheme wasn't worth the effort for a single filesystem given what could potentially break.
Either way, don't make foolish comments about development of an open source project dying. It's open source. Pick it up and maintain it yourself. - ropers, on 10/12/2007, -0/+2@xen0blue:
You're comparing apples and oranges here:
Microsoft's filesystem is a **de-facto** standard, but prior to that fab NTFS-3G effort (which, incidentally, just hit 1.0, cf. http://digg.com/linux_unix/NTFS_3G_has_just_gone_1_0_Pop_the_champaigne ) nobody really knew how NTFS worked. Expect for MS and those who signed in blood on the dotted line.
The vast majority of Linux' filesystems are **technical** standards. They're known and documented and in terms of implementing them, to quote Woz: "Everybody is welcome." - ropers, on 10/12/2007, -0/+2If you want simplicity, install OpenBSD. (www.openbsd.org)
It might not be the kind of simplicity that you're thinking of, but it's the kind of simplicity that matters even more than so-called "user-friendliness", cf. http://www.xml-dev.com/blog/index.php?action=viewtopic&id=149 .
/me zips flamesuit - ropers, on 10/12/2007, -0/+2@damentz:
It occasionally happens that heavy digg users post a comment in the wrong window/tab/textbox. Digg doesn't allow you to delete your comment, but it gives you a 2 minute grace period within which you can still edit your comment after posting it. Thus it is common practice for people who have misposted to replace their wrongly posted comment with a request to digg down that particular comment.
Clear as mud? - pingveno, on 10/12/2007, -0/+2Apparently the NameSys developers (NameSys is the company that makes the Reiser file system) intentionally put more of the load on the CPU than on the hard drive. The hard drive is the bottleneck, so shifting that weight to the CPU is more efficient overall.
I heard that they're working on a compression plugin as well (Reiser 4 has some sort of plugin architecture). Apparently that could go further in shifting weight away from hard drive reads/writes because less data must be read/written. - Teaboy, on 10/12/2007, -1/+2damentz: If you're reading more than one article at once and you need to login to comment then digg will redirect you to the last page you had open, and then you could comment on the wrong article.
- geronimo, on 10/12/2007, -1/+2I've seen 4,000 people on AFS and it worked just fine and it was secure with kerberos, I think what matters is what you do with it, it's not a speed daemon.
- emo1313, on 10/12/2007, -0/+1I gotta say... My not being a linux user, i clicked that link expecting some 3 page dissertation (each one) on files systems so dry that it made my martini look girl drink drunk... but I have to say, that's what I was looking for, smart, sharp and to the point...
- GMorgan, on 10/12/2007, -0/+1@duhblow7
I don't have a link but it appears via emerge as you pull in gentoo-sources (or at least it did). Their main complaint is about Reiser4 to be fair and they suggest a whole raft of traditional file systems instead. At the end it says something along the line of even Reiser3 wouldn't be totally insane. My thought on reading the message were that they didn't think much of R3 either but that it was way better than R4 ATM. - smellinator, on 10/12/2007, -2/+3Wouldn't it be cool if the Reiser "file" system came baked in a cake?
- daemonical, on 10/12/2007, -0/+1Most people are using ext3 on Servers and Workstation, because it's rock-stable - so maybe you do need lot of filesystems to hype the quantity but in the end they do not matter at all.
- Innatech, on 10/12/2007, -0/+1MS claims that NT is POSIX compliant (providing the POSIX subsystem/SFU is installed)--mostly to fulfill requirements for acquisition by some government entities. That is, of course, a joke--but less so since they replaced their own crappy POSIX subsystem with what used to be Interix, which they bought from Softway Systems and renamed "Services for Unix." The suckage was reduced, but no one really thinks NT is POSIX compliant.
A few top google hits:
http://www.phptr.com/articles/article.asp?p=691503&rl=1
http://www.microsoft.com/technet/prodtechnol/windows2000serv/reskit/prork/prdf_fls_ovte.mspx?mfr=true
http://support.microsoft.com/kb/308259
http://www.microsoft.com/technet/interopmigration/unix/sfu/default.mspx
http://www.microsoft.com/technet/interopmigration/unix/sfu/migun2wi.mspx - DrDabbles, on 10/12/2007, -1/+2ZFS isn't infinitely expandable. It simply uses 64bit addresses. Yes, I know about the "boiling the world's oceans" argument...but that doesn't make it infinite.
And XFS FTW. Love it. And xfs_fsr cleans up all those torrents you download and puts them in nice consecutive chunks on the drive. Very tidy. Very quick. But, grub _does_ do XFS. Your distro simply compiled grub without XFS support. Even SuSE can boot from XFS. - maninblac1, on 10/12/2007, -1/+2It's not the speed of the system, it's the bugs in the client. From transarc daemon errors to the inability for the client to work 100% of the time. It's really bad when you client asks for tech support, you visit the client, and either he reproduces the problem (which is usually resolved magically by reisntalling the software), or reruning the ticket manager just causes it to work again. Oh, and i should also add that the AFS context menu handler hangs when you don't have AFS connectivity, which is not required in our environment it's only active when AFS access is needed, which is a real pain for the users. The number of problems we've had with AFS is far greater than i want to get into.
What do i tell the user when he asks, "why do i need to reinstall the client" or "why do i need to reboot the machine", all i can say is this, AFS is crappy software. As far as OSS goes, AFS is probably the worst suite i've ever seen, no joke. We use the leash32 kerberos ticket manager and kerberos authentication as well, as a note. -
Show 51 - 72 of 72 discussions
Check out the new & improved 
© Digg Inc. 2009
— Content created and posted by Digg users is