Hacker News Reader

Comment by antongribok | Hacker News Reader

I actually recently switched from my laptop to my desktop as my main work machine, and due to some weird partition choices previously (long story), I temporarily ended up with my /home/<work_user> directory on a btrfs filesystem that's sitting on top of 3 Seagate Exos 20TB drives (instead of my main NVMe).

Hearing the drives has been really nice actually, and got me noticing all kinds of interesting and sometimes unexpected behavior going on with my system, and actually helped find a bug with my terminal multiplexer.

With 64GB of RAM my entire home directory fits, so only writes go to the drives, and it's been surprisingly performant for my workloads.

View on HN

On old PCs drive activity indicator LEDs (good) or drive mechanism sounds (better) were always a great way to build an intuitive feeling for what your computer was doing. People who got into PCs in the post-SSD world have no idea how the sounds act as a window into the computer's operation.

This kind of intuition goes back a long time, too. Levy's "Hackers" describes the MIT TX-0 having a CPU register connected a speaker and the hackers who programmed the machine being able to suss out how their program was doing by the sounds the machine was making. Indeed, that's closer to the "metal" of the machine than disk activity sounds.

The closest thing we have today are the fans throttling-up like jet engines, but by the time that happens things have usually gone well and truly off the rails and it's just an indication of "all hell is breaking loose w/ your CPU".

Yep, fan speed up is the only time I now think "wtf is my computer doing right now?" when I feel it was a common feeling before.

Large quiet rooms with only platter disk sounds to interrupt the silence were experiences that I quite cherish too.

I am all for building sonification tools. It is a long standing idea of our IT department and me to build some sort of university wide network traffic sonification tool. But today given the amounts of data and the speed involved you can only abstract it down.

So the major task will be drinking from the waterhose and having something useful fall out on the other side.

Growing up with those beige deeping beasts I always remember the activity indicators going wild. They also seemed to make more noise during HDD activity.

The drive mechanism sounds were they just louder back then or? You can put a HDD in a PC and it won't make the same grinding noises the old ones used to

Older hard disk drives had a more melodic sound to their seeking, to my ear. My first PC with a hard disk drive had an ST-251. That had a very distinctive, almost musical sound.

As lovely as 3x20TB RAID sounds, RAID on btrfs is never a particularly good idea. https://wiki.archlinux.org/title/btrfs#Multi-device_file_sys...

After 17 years of existance one would assume they have fixed all major flaws. Anyone knows why raid is still broken on btrfs? And what is the next fs after zfs and btrfs? There should be a major contender by now, but I see none.

I believe the recommendation is to run btrfs on top of mdraid, which works fine. Meta is said to use btrfs, but apparently they don't care about the raid implementation, so there's nobody paying for that development.

OpenZFS works fine, but it can be tricky to install due to the license thing, not all distributions support it in their installer.

Bcachefs is said to be the up and coming contender for both, but it was merged to the mainline kernel in January, so probably at the moment it is tricky to install as well. I would think many of this year's distribution releases will make it available.

IMO the "next fs" is just zfs. They somewhat recently merged RAIDZ expansion feature https://github.com/openzfs/zfs/pull/12225 and make regular improvements. If no file system has what you need today, zfs will probably be the first one to have it "tomorrow," imo.

> Anyone knows why raid is still broken on btrfs?

For all the talk about the raid5/6 issues on btrfs, people don't seem motivated enough to actually spend time fixing this. It's almost as if mdadm was enough and there wasn't that much drive to make it happen.

> And what is the next fs after zfs and btrfs?

http://bcachefs.org/

While BTRFS' RAID has issue, BTRFS on RAID can be done.

I do something like this, https://kb.synology.com/en-us/DSM/tutorial/What_was_the_RAID...

Basically, mdadm, then on top, format with BTRFS

Don't overdramatize, the link doesn't suppport "never". RAID 5 or 6 is not a good idea.

RAID 1 or 0 is just as a good idea as ever.

Not sure if I agree with you on RAID 0, but RAID 1 certainly is, and that's indeed what I'm running.

The 3x20TB btrfs setup I mentioned is configured with RAID1C3 for metadata, and RAID1 for data, and works just fine with even or odd number of drives.

It's funny how people assumed RAID5 when they saw 3 drives.

I switched to this after years of running on ZFS, and for my workloads btrfs is faster on Linux (not to mention the licensing/packaging mess).

I never said RAID0 was a good idea, just that it's as good as it ever was :P

Makes sense. I should have read your comment more carefully! :)

Raid 5/6 should not be used. Other levels work just fine. This is not a raid-in-general problem.

Reply on news.ycombinator.com