ReiserFS File-System Expected To Be Removed With Linux 6.13

[Mark Rose]

So if "no one is using it" is a justification to remove filesystems, that begs the question why we still have all stuff under "Misc filesystems", particular... when was the last time you used ADFS?

If ReiserFS were just floating around there would be no reason to remove It, but it has Y2038 problems and a maintenance burden. JFS (which is nice for its very low CPU usage) is similarly being removed for the same reasons.

[cjcox]

I'm going to go back into the past and point out that there was a time when reiserfs was the "easy" choice for a modern world that needed a fast growable filesystem in the enterprise space. Reiserfs, combined with Sistina's LVM (LVM1) was a marvel at the time.

Also, we learned about "better ways" of dealing with the entire kernel development team, again, regardless of your perspective of the "man" behind the curtain.

While it's been quite a long time since I used reiserfs, I'm very glad it was there, and super glad for SUSE's support efforts for it early on. The SUSE guys truly made reiserfs work. Can't understate that.

We also learned about filesystems and integrity (all types) because of reiserfs. Nuff said.

[fitzie]

I used ReiserFS back in the day before EXT3 and JFS were available. It's was one of the worst filesystems I used, only FS used under Linux to eat itself and also had a bad habit of getting slower and slower to the point of bootup taking many times times longer, the longer the filesystem was aged.

ReiserFS was made FAT32 under Windows 9x look robust.

Xiafs was the worst. mostly because of its users.

[vladpetric]

With software that is not properly maintained, things f.ing break ... all the f.ing time, especially when used at scale.

The primary filesystems (e.g. ext, xfs) that we use on Linux had their bugs ironed out by virtue of running on so many servers. With BTRFS we were lucky that Mason managed to bamboozle FB/Meta to use it at large scale, and they fixed a lot of bugs that way (not all of them of course ... RAID5/6 are broken at a design level, unfortunately).

What kind of deployment/workload scale did Reiserfs see over the last decade or so? Honestly I'm willing to guess it's less than 10k server years, which is really low of course.

Murder is of course abhorrent but using free software produced by a convicted murderer does not mean condoning murder.

[ayumu]

The elephant in the room: We are even talking about this because Linux is shit, there's no stable APIs for filesystems, and on top of that they run in supervisor mode for no good reason.

[mobadboy]

since i imagine they are looking for something to fill its gap, maybe they could git pull zfs

[erniv2]

Back in the days ReiserFS was the hot shit, if you had a platter based drive and listend xfs would go crrreeek crrreekkm ceeek, and ext would god creek creek creek , while reiser fs would go creeeeeeeeeeeeeeeeeeeee, yes back then you could hear your harddrives, nowdays all those sccreeching harddrives wish they where formated with reiserfs.

[Kemosabe]

Okay I am gonna do it because its probably the last chance to do it:

Yes its time to finally kill reiserfs, Just too bad reiser could not kill it himself.

What do you know about Reiser, what he did? Your comment is an epic expression of immaturity.

[Britoid]

Michael should have posted this article then turned off comments for it since this topic invites nothing but derisive & tasteless comments that add nothing of value ... except for being clickbait.

Please stop trying to kill the mood

[Britoid]

The elephant in the room: We are even talking about this because Linux is shit, there's no stable APIs for filesystems, and on top of that they run in supervisor mode for no good reason.

A filesystem driver has full access to the system regardless if its in userspace or kernel space, it can replace data from the filesystem being loaded by the kernel, including executables.