ReiserFS is now “obsolete” in the Linux kernel and should be gone by 2025 | Hacker News Reader

ReiserFS is now “obsolete” in the Linux kernel and should be gone by 2025

arstechnica.com

26 pointslisper9 comments

I wonder if ReiserFS would have suffered as much from the murder if it hadn't borne Reiser's name.

After all he was the lead dev but not the only person who worked on it. But I can imagine nobody wanted to take on a project with his name.

I was in the storage industry back when the unfortunate events happened (even knew an engineer who was personal friends with Reiser). My take is that:

1) ReiserFS had some innovative features, but was also known for shredding data; in fact I personally lost data to some kind of bug in the driver, not as a result of hardware failure. So it didn't have a great reputation among professionals.

2) Some of the really useful features were also provided by competing file systems like XFS and ZFS; also, EFS was getting better every day. Many sysops sidestepped to those (over a period of several years) and called it done.

3) A number of ReiserFS capabilities were focused on saving disk space at the expense of performance. (IIRC it was block packing and inode-only storage for small files, but it's been a while and I could be misremembering.) These were effectively obsoleted by the drive industry's polynomial increases in storage capacity.

4) AFAIK it was a one-man show, there wasn't a cadre of contributors and maintainers to carry it forward.

So combined with what you said, it wasn't compelling enough a project to resume development even with a rename.

> AFAIK it was a one-man show, there wasn't a cadre of contributors and maintainers to carry it forward.

Why does everyone forget Edward Shishkin? He was the other main developer and he's still working on it... he announced Reiser5 not long ago.

That's good to know, and I apologize to Mr. Shishkin! Reiser did have a flamboyance that would have made anybody not in the spotlight hard to see.

I don't remember specifics any more and don't have enough time to research at this moment - maybe later tonight.

Around the time when ReiserFS v4 came out I remember there being a lot of strife around performance numbers - specifically that nobody could replicate anything even close to the performance numbers that the ReiserFS team was claiming.

Between that and other interactions, there was an increasing divide between the ReiserFS team and the kernel community at large. I believe (but never confirmed) there was also a concern that the filesystem may go commercial which did not help matters.

While the murder didn't help matters, by time news broke there was already very little interest in ongoing community support. I suspect that even if Hans didn't murder his wife and merely abandoned ReiserFS, it would have suffered the same fate.

I didn't have much experience with ReiserFS 4, but had good experiences with ResierFS 3 on many devices over several years before moving on to better supported file systems.

Just out of curiosity, may I ask for which reason did you use ReiserFS v3 and on what kind of devices?

What reason? Nobody ever asked that.

Kinda funny, I was learning about file-system data recovery and forensics around 1999-2000 and was impressed with how much data that could be retrieved from slack-space [1] on hard drives. I had originally learned this on FAT based file systems. Wanting to expand my skills, I built a linux kernel with all the available file systems at the time, including ReiserFS and then created files systems to experiment with. I was surprised to find that there was no slack with ReiserFS. Without getting more long winded, that caused me to take a closer look at it. Put simply it got my attention.

By around 2001 I was using it on my personal workstations and occasional laptop. Eventually it even made it onto my personal co-located servers. At the time I had the delusional idea of making my own security-focused linux distribution, so I was re-installing my devices frequently. Probably why I never had an issue with failure.

The one exception was one of my co-located servers that I used for version control that was online with reiserfs for almost 2 years without issue. I migrated to a different server and rebuilt that one before it failed.

(edit) I almost forgot - Something I did really /love/ about it was that it handled filesystems with many (hundreds of thousands) of tiny files, symlinks, and directories without breaking a sweat. I was using it as a weird document/graph-oriented database for an app I wrote and was experimenting with (2002-ish?). Now I remember its architecture I'm surprised how well it worked considering I was too inexperienced to understand what indexes were let alone how to use them.

All other file systems I tried that with would run out of i-nodes and become incredibly slow at a mere fraction of what the reiserfs based version was able to achieve.

[1] - https://www.techtarget.com/whatis/definition/slack-space-fil...

I used to use ReiserFS exclusively once upon a time. That was because ext2 used to get corrupted easily when the power went out. Once ext3 and ext4 came out with journalling, I found that they were just as good as ReiserFS.

ReiserFS is nice but also unmaintained, something that doesn't seem is going to change. Decision makes sense.

Reply on news.ycombinator.com