Does /g/ use XFS?
No. XFS is extreme copium by redhat for wasting money on an inferior technology that can never be better than BTRFS or ZFS.
>>109389158Gentoo Handbook suggests XFS
>>109389192well they are wrong, using out of date info that BTRFS is not stable...EXT4 is actually more stable and recoverable than XFS. Fedora, which us uncucked RHEL, uses BTRFS by default.
>>109389136EXT4 and NTFS for dataF2FS on SSDs
>>109389536is f2fs really that better on ssds as of now? i know ssds have their own algorithms for storing and retrieving data optimally by default, is this not enough?
>>109389136>Does /g/ use XFS?No, I just use btrfs for the desktop and ZFS for servers.
>>109389136used it for a while, can't say I ever really noticed a difference compared to the usual ext4
>>109389548Last time I checked phoronix FS benchmarks was like 2019 or something. it was doing alright, the thing about it is that it's supposed to be performant while reducing wear due to the way it stores data. I believe samsung has been using it on all of their devices since 2018
>>109389536>>109389548F2FS has shit fsck. One hard shutdown without sync your data is gone most likely
>>109389591wow this sucks. reminds me of using windows on fat32. I wonder if this is still true today
yes, but wish I had gone with ext4 because now I want to shrink my partition
>>109389591well, I use 'commit=60' anyway, so it can't be worse.
Part of the reason I continue to use xfs is some custom hydrus-like software that I wrote years ago.Unlimited custom xattrs, yes I'll have that.
>>109389136I use it for home and storage and leave my system files with ext4. something like this:/ - ext4 512gb/home -xfs 513/extra -xfs 2tbI haven't notice anything off with xfs, seems nice and tight, I dont use any features it has but over nvme I doubt there's much.
>>109389136No I use btrfs.
>>109389136>not using butterfs>not having a tasty filesystem
I use xfs on my unraid nas. It very stable and comfy but lacks the bazillion tools btrfs has.
>>109390621>Unlimited custom xattrs,anon are you storing your booru tags as xattrs? thats amazing
>>109389136I use Btrfs like a chad.
Will Kent's AI waifu be able to charm Linus into putting bcachefs back in the kernel?
>>109389136>I boy soft Do Japs really
Obsolete by btrfs
>>109389136yes, on gentoosee >>109389192
Literally why would I use anything but ZFS if I wanted an alternative file system?
>>109389136Like to keep things simple withext4 on SSDbrtrfs on HDD
>>109394825Format overhead, both in terms of storage space and CPU wasted. Out of control fragmentation.
>>109389136LVM/XFS for boot drives always. Unless the boot drive is so slow it's actually faster with compression. Anyone who thinks enterprise larp or enospc fs is better doesn't test their ideas.
>>109389136my setup:rootfs xfsother drives(+mirror raid) btrfs w/ zstd compression
>>109390621I did this for a while and then accidentally 'backed up' my collection without xattrs. Tooling just doesn't respect user attrs even though there are actual defined namespaces for them.
>>109389576They use a mix of ext4 and f2fs, but it's also their pet project. The faster storage is the more it punishes CPU for trying to outsmart it.
>>109389591i've had literal power outages before i got a ups and i've moved the root across 2 f2fs formatted root partitions in the 7 years i've had this computer and the only time i lost any data was once when i accidentally wiped my rootfs (fortunately had a recent backup on hand bc i was switching drives)i also hard powered off many times and it didn't shit the bed even when i had a shitty adata xpg sx6000 litemy main complaint with f2fs is that you can't shrink it like you can with any other even remotely mature filesystem
>>109389136it's my understanding xfs still beats the alternatives when it comes to large filesizes and raw speed in general. that makes it specifically suited for spinning disks and storage drives.i use btrfs for root partitions and flash storage, xfs on hdds. some people will say ext4 but btrfs is a direct upgrade over it as far as i'm concerned.some people itt are saying f2fs but that's only intended for storage flash drives like sd/micro sds and usbs. in that case though it's still better to use exfat for compatibility. you might only use linux on your hardware but the one time you can't plug a usb drive into a chromeos or windows laptop will be a huge pain the ass. portable storage should be compatible before it's fast.
>>109397641yeah basically, samsung (and 90% of other android smartphone manufacturers) uses it for the userdata partition because that one's gonna be hit the hardest with reads/writes obviouslyand the rest of the system partitions are kept as ext4 (also most of them are read-only) due to ext4's reliability and maturity
>>109394878this makes it out like you'll kill your ssd if you use btrfs, that's not the case. for one, in most uses you'll replace an ssd because it's old/slow/small before you actually wear its' blocks out. and for two, the write amplification they do in that study you posted was random individually-synced 4k modifications in a 100M file. this is by far the worst-case for btrfs and is only realistic for very specific real-world applications (such as databases) and it's often recommended to turn off CoW for things like databases for this specific reason (you can turn off CoW in btrfs on a file-by-file basis)t. running btrfs root/home on an ssd with 5.5 years of power-on time and it's still 81% health
>>109398257Your understanding is retarded. Nobody cares how slow a filesystem processes on HDD. It's not a bottleneck. All these XFS benchmarks are on mid-high end flash, which is what it's best for. Along larger scale abstract block storage.
>>109394878Those are metadata figures. 33 * nothing is still nothing. Those filesystems are using extra metadata to skip writing actual data in its full size or more than once. Actual WA is LOWER.
>>109398257most people aren't deciding on a filesystem for a large, many-user, high-throughput file server. any differences in benchmarks won't be felt on your personal computer regardless of what filesystem you use, which is why if you're picking xfs over btrfs because it's "faster", you're probably doing it wrong. btrfs on my desktop isn't slow, i might be losing milliseconds here and there but nothing that can be felt. features like snapshots, compression, subvolumes and checksums are more than worth any imperceptible loss in performance. it's also not because btrfs is less efficient, rather features like checksums and cow cost performance in some workloads, just like how journalling can but you don't see people sticking to ext2 (or ext4 with journalling turned off which is a better choice if you do want to avoid journalling like say for a ramdisc since journalling makes no sense there) since most people understand journalling is worth the performance hit
