Reiser4 Benchmarked On Linux 3.5

[disi]

Most distributions do not have first party support yet, although the number has been increasing. 5 distributions have first party support as of this year. That is up from 0 last year.

Hi ryao,

I got fed up with FreeBSD and support for certain SATA controllers, so I whopped the OS from the disk yesterday and put Gentoo on it.
1. The Gentoo LiveCD 12.1 with ZFSOnLinux opened my existing pools without problems (once created with FreeBSD)
2. install went fine (I didn't use ZFS as rootfs but ext4)
3. all pools are available during boot and much better HW support with 3.4 Linux kernel

[droste]

so why we don't use ZFS? (don't answer this was only a rhetorical question to show you how stupid it is to write about ZFS in this tropic.)

Q, it is "topic" not "tropic". Sorry for being OT, but you make this mistake all the time. And it also shows that you're Q, which you tried to hide so hard ;-)

[ryao]

so why we don't use ZFS? (don't answer this was only a rhetorical question to show you how stupid it is to write about ZFS in this tropic.)

Most distributions do not have first party support yet, although the number has been increasing. 5 distributions have first party support as of this year. That is up from 0 last year.

[WorBlux]

Typically murders mail order wives and makes shitty defenses in court then gets beaten up badly in prison. Also, even if Hans Reiser didn't murder anyone, the kernel people NEVER liked him. We would have never seen Reiser4 becoming mainstream anyway, despite the fact that ReiserFS was so damn popular. I have a hunch, if it was developed more, it would be far ahead of any Linux filesystem speedwise (for desktop loads at least.)

From what I've been able to gather he wanted to do several things that just didn't work/ broke compatibility with the traditional Linux and UNIX filesystems.

Here's a taste of some of it.

http://yarchive.net/comp/linux/reiser4.html

[necro-lover]

ZFS should be able to outperform it.
ZFS should be immune to this issue.

so why we don't use ZFS? (don't answer this was only a rhetorical question to show you how stupid it is to write about ZFS in this tropic.)

[ryao]

Typically murders mail order wives and makes shitty defenses in court then gets beaten up badly in prison. Also, even if Hans Reiser didn't murder anyone, the kernel people NEVER liked him. We would have never seen Reiser4 becoming mainstream anyway, despite the fact that ReiserFS was so damn popular. I have a hunch, if it was developed more, it would be far ahead of any Linux filesystem speedwise (for desktop loads at least.)

ZFS should be able to outperform it.

'whooped' in a 'majority'? Really? Well technically - because of threaded io tester.... if you take that one out everything changes. Or you could have written 'btrfs is whooped' ... well... whatever.

Btw, why did you miss to mention THE r4 feature? atomicy?

Rename data loss can happen with every FS - except reiser4.

Or wandering journal and dancing trees which are a nice way to spread the write load over the disk?

ZFS should be immune to this issue.

[garegin]

ssds are gonna become the standard in five years anyway.

[stan]

BTRFS sucks

Sorry to say it, but what's all the excitement about BTRFS when it sucks so much compared to EXT4. Why would any desktop Linux user in their right mind chose BTRFS???

[Adarion]

They never let it in the kernel. I personally believe that some ext people were even scared of ReiserFS's efficiency and didn't want it in the kernel. Not monocausal but this was also a reason.
Now, for being developed by so few (one?) persons, and being out of tree it still makes a good run. It's sad that it never really had the chance to go mainline. Ext was never _that_ great and btrfs is still fiddling with some issues and still not announced as ready for productive systems (but I'm glad Chris Mason is not at Oracle anymore).
But so there are some things, people, good ideas, sometimes ahead of their time. Masses laught at them, but much later everybody uses the same principle and agrees that this is now common standard.

[energyman]

'whooped' in a 'majority'? Really? Well technically - because of threaded io tester.... if you take that one out everything changes. Or you could have written 'btrfs is whooped' ... well... whatever.

Btw, why did you miss to mention THE r4 feature? atomicy?

Rename data loss can happen with every FS - except reiser4.

Or wandering journal and dancing trees which are a nice way to spread the write load over the disk?