Linux kernel edges closer to dropping ReiserFS

Linux kernel developer Matthew Wilcox has proposed removing ReiserFS from the Linux kernel – a relatively rare step. This isn't due to a problem or bug in ReiserFS: it still works fine and has plenty of features. The problem was that ReiserFS code in the kernel used some API calls that nothing else did, preventing them from …

[JessicaRabbit]

Not sure 'just' has anything to do with it (and nor should it). It's clear that the proposal is for entirely technical reasons and the suggestion to mark it as deprecated for a few years before actually removing it makes good sense to me at least.

[Hubert Cumberdale]

Quite. I see no "moral" reason to remove it from the kernel: there are plenty of examples in the art world of people the deeds of whom we seem able to separate quite happily from what they created.* The technical and the practical should be the only issues here. After all, people still seem to love the VW Beetle...

^{*Although weirdly, we don't seem to want to do that with Rolf Harris.}

[John Savard]

Puzzled

Surely if ReiserFS is useful, it can be forked to a version that has nothing to do with its original developer.

[thames]

Re: Puzzled

The issue has nothing to do with who the original developer was. It has to do with that it had few users to begin with (mainly Suse), and hardly anyone has used it in years. It was a niche file system to begin with and it's been an obsolete file system for years.

Nearly everybody uses Ext4, which does everything that they need. Other alternatives include XFS, ZFS, and BTRFS. ReiserFS was always an also-ran a long way behind the leaders.

As I understand it, Hans Reiser ran a small company which licensed out a non-open source version of his file system to proprietary SAN vendors. The open source Linux version of ReiserFS gave his small company a larger user base which gave the file system more testing than his small company could have economically done themselves. It always however had very limited use as compared the the more mainstream Linux file systems.

If anything what this shows is the overall stability of the underpinnings of Linux, that ReiserFS should have continued on for so long with so few Linux users.

[rcxb1]

Re: Puzzled

> ReiserFS was always an also-ran a long way behind the leaders.

Not true. It was quite popular when it first came out, right up there with ext3. It was rather common right up until Hans' arrest (whether or not that event was part of the cause). At the time, XFS was immature and early adopters were reporting data loss after power failures. Ext3 was a good upgrade over Ext2, but performance was not great. Ext4 was years off, but it greatly improved matters.

I stuck with Ext3 for the backwards compatibility myself, only briefly trying out ReiserFS to see what the hype was about. It did help with video playback, as you didn't have the stutter on Ext3's sync every 5 seconds (nrsec)...

That was nearly two decades ago, now. ReiserFS hasn't improved enough over the years to keep pace with others, so there's no reason at all to use it any longer, and removal seems overdue.

[Ozan]

Re: Puzzled

I was using ReiserFS at that time as well. Basically, Version 4 didnt make it to the kernel for lack of support (REiser's company was gone and he was in jail. Not enough maintainership behind it. It is still same.) And Version 3 is quite old and has basic maintainership right now.

It is unfortunate, Version 4 was quite a step-up.

[Joe W]

Re: Puzzled

We got burned by one of the earlier versions, leading to the name "ReißwolfFS" (shredder FS) for it, I think v1. Don't use anything in v1, wait at least until v1.2 or better v2.1.

Reiser3 is still used in one of our systems - there was no reason for us to change to a different FS in the past decade and a bit. At that time, there were issues with ext (which I cannot remember) we did not like, and XFS was rubbish, and ZFS did exist, but only on Solaris.

Bleargh, I hate having to migrate all the stuff to a new disk. Though maybe the system might be due to a completely fresh install, removing a ton of old cruft. Only problem is that this takes a couple of days time that I don't have - there are a few special configs for the web- and mailserver part that need to be tested thoroughly.

And those who pooh-pooh the space saving, saying it doesn't matter: care to buy me my new hard disks? No? Thought so.

[Ozan]

Re: Puzzled

First ReiserFS was indeed problematic. I dont remember too much indetail. It's been long time. Even Reiser is due to parole hearing.

[MacroRodent]

Re: Puzzled

The only problem is that if removed, newer kernels cannot access old partitions with ReiserFS. But with enough notification about the deprecation, those with important data on ReiserFS hopefully remember to migrate it.

[Gaius]

Re: Puzzled

> At the time, XFS was immature

At the time the Linux XFS implementation was immature yes, but the original XFS was (and is) rock solid

[crayon]

Re: Puzzled

"At the time, XFS was immature and early adopters were reporting data loss after power failures."

For me, after having twice suffered corrupted file systems using ext2/3, I changed over to ReiserFS and haven't had any problems since. That was years ago, nowadays I don't think I have any disks using ReiserFS anymore.

[FeepingCreature]

Re: Puzzled

I was using ReiserFS with Gentoo. Like any source-based distro, it involved making lots of small files that were written once, read once, then deleted. ReiserFS was amazing at that, noticeably faster than the competing filesystems at the time.

[DrXym]

Re: Puzzled

ReiserFS was awesome back in the day. I had ext2 partitions that could take 30 minutes to check whereas a journaled filesystem would do it in seconds. Anyway I think the reasons for using ReiserFS evaporated as ext3 and ext4 appeared allowing a forward migration path and most of the same benefits.

[katrinab]

Re: Puzzled

The problem isn't the name "ReiserFS". The problem is that for most people, the answer to the question "is it useful", is No.

I used to use it up until about 12 years ago, as it was a lot better than ext3. But when FreeBSD 8 came out, I switched to it, and to zfs. zfs is just sooooooooo much better than anything else out there. It is available on Linux now, and I can't think of any reason to use anything else.

[Doctor Syntax]

Re: Puzzled

Agreed. I also used it a long time ago but switched to ext3 & 4. I'd guess its main use now is probably on boxes that haven't been updated for a long time.

[guyr]

Re: Puzzled

Same, I used it for a bunch of years in the early 2000s. At it's height of popularity, ReiserFS was great compared to the alternatives that were available. Then it stopped being maintained. I switched to BTRFS once that got past version 1.0.

[Nate Amsden]

Re: Puzzled

Reiser3 certainly has some good use cases. I use it still today for those cases. Mainly situations where there are tons of small files. For my personal mail server I use it for my Cyrus IMAP mail spool which has about 1 million files in it(38G of disk space used). ext4 and others can be used as well they just consume more disk space because of all of the files.

Other use cases I use it for at work are for deployment servers that involve a lot of source code checkouts, again significant space savings(25-35% I think) for little or no overhead. ZFS with compression certainly can save a bunch of space in this situation as well however it has significant overhead both memory and CPU (especially considering ~70% of the VMs I have deployed are 1 vCPU). I do use ZFS with compression in some other use cases, and XFS in some cases as well, ext4 remains the only filesystem I use for the root filesystems.

Will be unfortunate if the day comes where reiserfs isn't available anymore. I never tried rieser4, last I heard I think it never got completed.

[Dave 126]

I remember reading about ReiserFS before any news of its creator, and it was in the context of next gen files systems. I got the impression at the time that there were limitations with ext3, and that people were looking at the future. BTRFS and ZFS would be discussed in the same articles. This was roughly the same time as Apple pulled some ZFS features into whatever they call their FS.

Since then, I've only seen mention of BTRFS on its journey to being recommended for production use.

[Gene Cash]

"one of the other ReiserFS developers"

So just exactly how many are there? One? Two? Half dozen? Any particular reason they aren't developing, other than they lost interest?

Not beating up on them... just would genuinely like to find out.

[aerogems]

I wonder

Say someone just did a find and replace, changing all references to "Reiser" to something else, and then submitted it as a brand new patch, if anyone would really know the difference.

Obviously murdering a spouse is wrong, but it also doesn't have any impact on the quality of the code. I have no horse in this race, it just seems like if the issue has to do with the fact that the guy who made it is an admitted murderer, they could just change the name and all references to the original author. If there's some other issue with the code or design of the filesystem then that's another story, but there doesn't seem to be any mention of that.

[katrinab]

Re: I wonder

But what USP does it have that would make you want to switch to it from whatever you are using at the moment?

[doublelayer]

Re: I wonder

You misinterpret the reason for dropping it. If they had a problem with the name or the developer, they'd have dropped it a decade ago. It's because it hasn't been updated in a long time, isn't much used, and may slow the development of other features that they're considering dropping it.

[mathew42]

Small (<4KiB) file size performance

My understanding was that one of the advantages of ReiserFS was performance and storage using tail packing for files were under 4KiB. Back 2000, this was a big deal, now not so much.

Now storage is cheaper and I've moved on to ZFS.

[Claverhouse]

I always just stick with Ext 4...

Reiser initially maintained his innocence, then changed his plea, confessed, and led police to his wife's body. The result of his confession was his sentence being reduced from a minimum of 25 years to 15, meaning that he may soon be eligible for parole.

.

Unlike the real bad men who steal from the Rich, who get 150 years; or reveal U.S. war crimes, who get 175 years.

[Fred Daggy]

It's no longer a killer feature?

[zuckzuckgo]

Clearly your just trying to get a Reise out of readers.

[Dave 126]

Oh for FS!

[Pfornistan]

A shame. I've used the ReiserFS since 2000, because I needed the stability of it knowing that because where I live, out in the sticks, brown-outs and black-outs and constant surges, etc, are common every day occurrences.

Using ext3/4 would be 'okay' for a few of these happenings, but eventually I would have a hard drive with nothing but a mess and I would have to either re-install or buy a new hard drive.

ReiserFS has saved my data and hard drives *innumerably* these past 22 years. I've *still* got 120GB hard drives sitting on shelves with dust on them that are >12 years old that I could pop into one of my current systems (14.2 or 15) and run that have survived this long *because* ReiserFS was the one file system that could handle the garbage of electrical problems we have in this area. I *still* use it over any other FS because it's tried, tested severely and true as far as I'm concerned.

I find it depressing that Linux 'experts' are going the way of M$ and 'fixing what ain't broke'.

[WolfFan]

Errm… if I had that level of electric power problems, I would invest in a proper UPS. And a generator. Which, in fact, is what I have actually done. The office has a UPS wired into the circuit box, so that certain outlets (fitted with red plates) are ‘computer power only’ outlets; after some efforts we managed to convince staff to NOT plug coffee machines, floor polishers, and such into them. It required firing a certain person who absolutely would not plug the coffee maker into anything other than a red plate outlet, but the Word got out. We have a generator with fuel for 3 days ready to go. At home I have a few desktop type UPSes and a generator with fuel for a few hours, no special circuits. The office generator starts automatically, the home generator lives in its box until needed. (No, there’s no fuel in the home generator, that is stored next to the lawnmower.)

[JamesTGrant]

Would 100% use the UPS backed coffee machine

In a power cut I’d have thought a nice cup of coffee was page 1, front and center, if the business continuity manual. I’d even spec the UPS for it!

(Typical server spec 400watts, need to keep it running for 1hour? Coffee machine, say 1kW for 5min? Cheapy UPS can wheeze out 6A (just about) so, provided you don’t mind no network or monitor, you’ll enjoy a nice coffee in the dark whilst the server room gets nice and toasty.

[Hubert Cumberdale]

Re: Would 100% use the UPS backed coffee machine

That's why I keep a Coleman stove and a stack of fuel bottles at the ready.

[Henry Wertz 1]

Is it maintained

I did use reiserfs once, I found the performance to be quite good and had no problems with it that I recall. But subsequent systems I went back to ext3 then ext4, generally just out of laziness of going with the default (ext4 is no slouch.)

I guess it is a matter of, how extensive of changes are going to be made to the kernel filesystem API? If it's a flag or two, and they're already getting patches to avoid having to use them, no big deal. If extensive changes are planned, and they would require a lot of work with reiserfs, I guess it's a matter of if patches show up in a timely matter or not. If so, reiserfs is presumably supported and I'd keep it in. If not, it's not getting adequate support and I could see dropping it, it's not good to have a single filesystem holding back development.

It would be interesting, there's all these hardware surveys (CPU make & model, GPU make & model, RAM, storage capacity, etc.) but it would be interesting to have one for filesystem type. I wouldn't be surprised to see reisefs has very low usage, but I really don't know I was surprised to read some NAS model was using btrfs.