|
Mr Dog posted:Actually btrfs is, in fact, bad, and violates the end to end principle. I told Ubuntu and Fedora to use btrfs and haven't had any problems with it. Other than the fact that when I try to make it mount the fs and use compression ubuntu says "well the SUPERBLOCK doesn't say to use compression gently caress you" and ignores it
|
#
?
Apr 9, 2016 18:56
|
|
|
|
| # ? Apr 29, 2024 14:22 |
|
|
it's a copy-on-write fs that completely shits the bed performance-wise if you try to store anything nontrivial on it like a database or (ironically) a systemd binary journal (systemd has a btrfs-specific workaround in it for this problem). zfs has similar design goals but none of btrfs' performance problems. btrfs also has a nasty habit of suddenly running out of disk space despite claiming that you have plenty of disk space free. they might have fixed this recently idk. also compression and encryption are better handled at the application layer, for the same reasons as in computer networking. to give one example, LZMA and H.264 are both types of "compression". which is appropriate depends on the application and the expected access patterns. but lennart poettering has a huge boner for btrfs so he's going force people to use it anyway.
|
|
#
?
Apr 9, 2016 19:22
|
|
|
doesn't zfs work best with ecc ram and stuff? that would make it a total non starter for anything except linux on the server/workstation
|
|
#
?
Apr 9, 2016 19:27
|
|
I CANNOT EJACULATE WITHOUT SEEING NATIVE AMERICANS BRUTALISED!
|
probably, idk. the btrfs and zfs merkle tree integrity stuff is something that shouldn't have to exist. hard disks and ssds already stuff your blocks with error correcting codes and return block io errors if those eccs fail. of course, if the manufacturer cheaps out on the ecc redundancy in order to squeeze more capacity out of the storage medium then yeah you're going to get undetectable bit errors. i got into an argument about zfs with a friend of mine recently. he bought some ultra lovely seagate drives that fail every three months or something, but he says he doesn't care because he's from europe and seagate are obligated to keep replacing them for five years, or something like that. so he makes heavy use of zfs raid and integrity checking. i told him that he might value a drive that's known to corrupt his data at $0, but i value such a drive at -$infinity, and that his pathological setup is really not an argument in favour of zfs. basically i'm a big fan of composable abstractions, and block-device-with-DISCARD is a nice one that mostly does the job. like, i'm not going to get religious about it, but anything that breaks layers as egregiously as btrfs does smells really bad to me and it needs to demonstrate a very clear win for me to accept it.
|
|
#
?
Apr 9, 2016 19:33
|
|
|
What's wrong with ext4
|
|
#
?
Apr 9, 2016 19:34
|
|
|
nothing at all, it's what i use. then there's xfs which is, i dunno, basically the same thing as ext4 except slightly better? i'm sure that's fine too. i used xfs briefly back in the linux 2.4 days. it would reliably zero out every single file that was open at the time of a system crash. every single time. suffice it to say i didn't touch xfs again after that. then again, linux itself pre 2.6 was a toy and you would have been a fool to trust it with anything important.
|
|
#
?
Apr 9, 2016 19:37
|
|
|
use hfs+, problem solved
|
|
#
?
Apr 9, 2016 19:38
|
|
|
blowfish posted:doesn't zfs work best with ecc ram and stuff? that would make it a total non starter for anything except linux on the server/workstation a total non starter for anything except the things linux is suitable for? oh dear that does sound bad
|
|
#
?
Apr 9, 2016 19:40
|
|
|
Mr Dog posted:it would reliably zero out every single file that was open at the time of a system crash. every single time. sounds like a great way to ensure that the integrity of your data is intact. no data? it was probably busted anyway
|
|
#
?
Apr 9, 2016 19:50
|
|
|
also fs compression rules because if you have a semi-modern processor instead of a Cyrix or whatever it makes computer faster. btrfs supports both LZO and LZMA so you can choose LZO for low overhead if you're not entirely storing compressible data
|
|
#
?
Apr 9, 2016 19:53
|
|
|
Soricidus posted:a total non starter for anything except the things linux is suitable for? oh dear that does sound bad Pardon me Sir, I use Linux on my gaming rig.
|
|
#
?
Apr 9, 2016 21:11
|
|
|
Captain Foo posted:What's wrong with ext4
|
|
#
?
Apr 9, 2016 21:18
|
|
|
Smythe posted:Pardon me Sir, I use Linux on my gaming rig. i'll count it as a workstation
|
|
#
?
Apr 9, 2016 21:38
|
|
|
Mr Dog posted:also compression and encryption are better handled at the application layer, look at this terrible opinion
|
|
#
?
Apr 9, 2016 22:10
|
|
|
Smythe posted:Pardon me Sir, I use Linux on my gaming rig. the cool part about falling out of gaming and having a beefy rig is that you can run an insane amount of virtual machines when needed. if dark souls was on linux i wouldn't have a windows partition.
|
|
#
?
Apr 9, 2016 22:40
|
|
|
b0red posted:the cool part about falling out of gaming and having a beefy rig is that you can run an insane amount of virtual machines when needed. powerful beefy rigs are so good dude. got mad poo poo in this beast and it just owns hard as hell.
|
|
#
?
Apr 9, 2016 23:25
|
|
|
the beefy rig does all computer tasks fast as hell and easily. loving around on some piddly laptop? gently caress off.
|
|
#
?
Apr 9, 2016 23:26
|
|
|
btrfs:zfs::android:ios
|
|
#
?
Apr 9, 2016 23:33
|
|
|
actually thats too generous, its more like firefox os
|
|
#
?
Apr 9, 2016 23:33
|
|
|
zfs can break my layers all day everyday i dont even need a safeword i give myself over fully
|
|
#
?
Apr 9, 2016 23:34
|
|
|
Smythe posted:the beefy rig does all computer tasks fast as hell and easily. loving around on some piddly laptop? gently caress off. wireless drivers? real men have beefcake rigs with direct lines. none of this pussy rear end wifi poo poo.
|
|
#
?
Apr 9, 2016 23:37
|
|
|
b0red posted:wireless drivers? real men have beefcake rigs with direct lines. none of this pussy rear end wifi poo poo. agreed
|
|
#
?
Apr 9, 2016 23:37
|
|
|
do Nvidia's Tesla cards have a different driver than the video cards
|
|
#
?
Apr 9, 2016 23:40
|
|
|
they provide it as a separate binary but since the version #s are same i always assumed it was the same driver with different device info/packaging
|
|
#
?
Apr 9, 2016 23:44
|
|
|
Mr Dog posted:nothing at all, it's what i use. Phoenixan posted:it's a working file system and we can't have something that works in the linux thread Okay this is what I thought
|
|
#
?
Apr 10, 2016 04:21
|
|
|
I used
|
|
#
?
Apr 10, 2016 06:06
|
|
|
why?
|
|
#
?
Apr 10, 2016 06:12
|
|
|
He hadn't murdered anybody back then and it was better than ext3 iirc.
|
|
#
?
Apr 10, 2016 06:13
|
|
|
ratbert90 posted:I used
|
|
#
?
Apr 10, 2016 06:38
|
|
|
ratbert90 posted:I used the butthurt the murderFS redirect generated in wikipedia article comments was glorious
|
|
#
?
Apr 10, 2016 16:08
|
|
|
also, someone should make yospoFS
|
|
#
?
Apr 10, 2016 16:08
|
|
|
fat already exists tho?
|
|
#
?
Apr 10, 2016 16:40
|
|
|
I installed reiserfs on the postop
|
|
#
?
Apr 10, 2016 17:00
|
|
|
Soricidus posted:fat already exists tho?
|
|
#
?
Apr 10, 2016 17:21
|
|
|
Soricidus posted:fat already exists tho? 
|
|
#
?
Apr 10, 2016 17:21
|
|
|
Soricidus posted:fat already exists tho?
|
|
#
?
Apr 10, 2016 17:23
|
|
|
Soricidus posted:fat already exists tho?
|
|
#
?
Apr 10, 2016 18:15
|
|
|
There should be a murderFS group tag.
|
|
#
?
Apr 10, 2016 18:15
|
|
|
a murderfsd-out gaming rig
|
|
#
?
Apr 10, 2016 18:54
|
|
|
|
| # ? Apr 29, 2024 14:22 |
|
|
Captain Foo posted:a murderfsd-out gaming rig find me in my garage on a sunday with my case door open, fans blazin, 20 htops open, screwdriver in hand. Just doing a little tuning before the week starts, gotta have my baby in top shape for the work week.
|
|
#
?
Apr 10, 2016 19:14
|
|