Few/scarse fs choice for freebsd

[sysman]

Why freebsd does not have same or similar linux hot develpment about filesystems? I think that is very bad for freebsd, 1:1 compatibility among differents (and widespread) filesystems is the key for the OS, to get more visibility over soho or desktop scenario. actually, freebsd has very tiny compatibility with different fs, I mean btrfs, ext3/4 (with no headache tricks, xfs, reiserfs ... I really hope this will be different with next bsd releases.

SYS

 

[SirDice]

Because UFS2 is good. Why change things that work?

As for the other filesystems, keep in mind that most (if not all) are GPL'ed. Which means they cannot be added to the FreeBSD base OS. The only way to add them would be to rewrite them from scratch. This will take time and effort. Effort I very much prefer to see being spent on FreeBSD itself.

 

[gilinko]

Btrfs is an attempt to write a version of ZFS in GPL, and as ZFS is already in the freebsd core that's not a problem. As for the ext{2,3,4} systems they are not very good compared to UFS which is why even Linus has mention on more than one occasion that the Linux kernel is still missing a good "native" file system. The rest of the journaled systems (jfs,xfs,reiserfs) is not even out of the beta stage in terms of a stable and reliable FS. That is for the Linux versions of the FS. Then we have the FAT/NTFS which can be used. And then there is the HAMMER file system among a lot of other either good or distributed file systems(HAST comes to mind).

But mostly the main FS of a computer is seldom shared with another OS, so that brings us back to technologies like AFS, NFS, SMB etc, which are present.

 

[sysman]

I'm not agree with you at all. I dont want to say UFS is bad, I think is really does good work, fat32 does it too (with limitations of course) or ntfs, but I think that a OS has to enhances itself on filesystems scenario in order to have a "flexible" OS environment. Ever lot of people I see ask for greater compatibility among systems, as such simple PCs or embedded systems, and so on... Here the main issue is not CHANGE but ADD new features and choices. Otherwise, freebsd (or other os too) could go on siding, this is equal to death. Of course, I prefer to see bigger efforts spent on freebsd itself too, but I guess there are lot of things to do (linux docet :stud), for example (and for me) improve browser plugins, java env, emulation with microsft world (still lacking), package upgrade management (actually powerful, but currently with some limitations), or compatibility with other hot sw like Truecrypt (currently I have no possibilities to switchoff my win/linux servers because they use Trucrypt fs management, and this sw is still the best to use). That's all, this is my modest opinion.

 

[Beastie]

I'm not agree with you at all. I dont want to say UFS is bad, I think is really does good work, fat32 does it too (with limitations of course) or ntfs, but I think that a OS has to enhances itself on filesystems scenario in order to have a "flexible" OS environment.

I don't understand the point you're trying to make. What FS FreeBSD doesn't support right now that it should support? What should it support them for (file storage vs. booting off of them)? Why should it support them/what are the advantages over UFS2 and all the FS FreeBSD already supports?
Can you be more precise please.

Ever lot of people I see ask for greater compatibility among systems, as such simple PCs or embedded systems, and so on...

Do *you* have compatibility problems related to FS? What are they?

Here the main issue is not CHANGE but ADD new features and choices. Otherwise, freebsd (or other os too) could go on siding, this is equal to death.

Are you talking about UFS2? Features are added from time to time, such as the new soft-updates/journaling system.

improve browser plugins, java env, emulation with microsft world (still lacking)

What do these have to do with the FreeBSD project?

 

[phoenix]

Think of it this way: which is better? 1000 different filesystems all in different stages of develpment, all with different features, none of which anyone can agree to standardise on? Or, a small handful of well-developed, well-tested, very well supported filesystems that can handle the majority of all file/volume handling duties?

Yes, Linux has a bazillion different filesystems that it supports ... and yet there is not a single filesystem that is good enough to known as the "native Linux filesystem". Maybe ext4 will become that fs. Maybe btrfs will. Neither are there yet.

Between UFS and ZFS, what can't you do, that requires the use of some other filesystem?

 

[expl]

... but I guess there are lot of things to do (linux docet :stud), for example (and for me) improve browser plugins, java env, emulation with microsft world (still lacking)... or compatibility with other hot sw like Truecrypt (currently I have no possibilities to switchoff my win/linux servers because they use Trucrypt fs management, and this sw is still the best to use)...

What do browser plugins have to do with the OS? Go write mails to the people who write those plugins.

Java env? People still use java outside schools?

How is "microsoft world" emulation is "still lacking" compared to linux? It uses almost exactly the same WINE code, FreeBSD even runs more smooth due to OSS since ALSA is bit bugged on WINE.

Just why Truecrypt is still the best to use? I can tell you without even testing that FreeBSD geli disk encryption will be way more faster an efficient than Truecrypt on linux simply because Truecrypt uses FUSE.

 

[fronclynne]

the journaled systems (jfs,xfs,reiserfs) is not even out of the beta stage

Not only that, but they have actual paid developers from rather large companies (in the case of jfs & xfs) trying to build, design, & maintain those piles of garbage. I dunno what to say about reiser, except that you'd better not talk back to ol' Hans (if he ever gets out).

 

[trasz@]

Basically, the Linux situation is just a waste of time. It's better to have the features in one filesystem, than to have a miriad of filesystems, each having one important feature and lacking others. For example (looking at the recent LWN.net news), there is a new filesystem, called next3, which offers snapshots. In FreeBSD, you simply have snapshots in UFS (and, obviously, in ZFS). From the user point of view, this means you don't have to mess with your machine changing filesystems. From the developer point of view, this means you have less code to mantain (and less duplicated code). The only significant disadvantage is that you cannot call yourself a "creator of a new filesystem" in your CV.

 

[fairy]

Think of it this way: which is better? 1000 different filesystems all in different stages of development, all with different features, none of which anyone can agree to standardise on? Or, a small handful of well-developed, well-tested, very well supported filesystems that can handle the majority of all file/volume handling duties?

For those supported outside of BSD projects there is puffs/refuse and fuse4bsd. It's better than scattering already scarce resources on new filesystems.

And implementing btrfs/xfs/whatever for fuse would benefit more users than just those few on freebsd.

 

[gore]

I agree that the person who posted this seemed to be a little misinformed about the BSD world at how it works with File Systems, but why would anyone say XFS is garbage? I can't even imagine saying something SGI did was bad... Yes, I'm partial.. And Biased. I'm OK with that.

Anyway, for BSD stuff, UFS and all of them in use on FreeBSD right now, ALREADY have the features that the Linux file systems have been trying to do. I have NO idea why ANY person would want to have Ext3 as their BSD File System.... Even if Ext3 was working great on FreeBSD I'd be sure to not pick it. It's NOT a great file system, and I've had to use it for a WHILE now.

I'm one of the people who uses both Linux and BSD very much. I don't pick sides in this either because they both are great to me which is why I have no problems saying that they both have great uses in everything. Right now I have an FTP server running Slackware, and was very happy to see a few of the BSD people saying to me they actually liked Slackware and that it was one of the few Linux distros they actually liked. I generally don't take sides in Linux VS BSD anything because I use both and like both. But I'm weird like that, I also use Emacs and Vi Both! I do like Vim more but that's for another reason as I do also use Emacs a lot. All of those things where people said something was so much better that they started Holy wars over it, please don't make it happen with File systems.

I'm opinionated enough to choose sides when I do think something works better outright, like how I hate Gentoo... But for File Systems, Linux has a few good ones that work just fine, and FreeBSD has the ones it requires to work great too.

FreeBSD's main File System with softupdates is something Linux just can't compete with in that way, it just doesn't need to use 12 when 2 will do. The reason I don't pick sides is because a few Linux File Systems, are actually nice, but FreeBSD's File Systems, do exactly what they NEED to do, and work great, so there is no point to change them around.

I don't have much experience in ZFS so I'm not going to talk about that but from what I've READ about it, it's amazing.

This is something I thought up years ago, and I think it's a good way to pick your software, File Systems, and Operating Systems:

If you were having Kidney Dialysis, and you had to choose which OS and File System and things that were going to run the machine doing that, would you STILL stand by your OS of choice? Or File System? (Thank you Marshall Kirk for talking about that lol).

I've asked that question to Windows Admins in the past, and they don't like it one bit. All the "Stop bashing Microsoft" sentiment in the WORLD will fly out the "Windows" when that question comes out lol.

And in case you're wondering, I've yet to find even one person who said they'd let Windows NT or 2000 or XP or any other Windows box run their Kidney functions lol.

 

[fronclynne]

I agree that the person who posted this seemed to be a little misinformed about the BSD world at how it works with File Systems, but why would anyone say XFS is garbage? I can't even imagine saying something SGI did was bad... Yes, I'm partial.. And Biased. I'm OK with that.

Typical hyperbole on my part. I'm sure if I had to use it day-to-day I'd be a lot more caustic.

I don't have much experience in ZFS so I'm not going to talk about that but from what I've READ about it, it's amazing.

I'm glad we have it, and for large, secure data pools it is probably very excellent*. Its resource use is a tad high for my tastes (read: utterly insane) and it won't make any applications I use run faster or better, so I'm also extremely glad I don't have to jump through crazy hoops to not use it (I'm glaring at you, OpenSolaris).



*If I had to guarantee to someone that their precious little bits would not be lost, it'd probably be my first and last choice. Otherwise, no.