Just after the second time of X.org locking up due to NVidia and hard-resetting, my whole XFS partition basically fucked itself up.
Screaming about some missing "magic superblock" and eternally finding a replacement (and failing).
Fuck you Holla Forums, And fuck you too, XFS.
...
...
...
you got memed son. use tfs: github.com
Dunno about you guys, but I have my money on NILFS. Dat continous snapshotting log structured goodness.
...
Just what shitty distro are you even using ? I've been on xfs for like a year and never had problems.
GILFFS
How can you fuck it up so bad? I'm on gentoo and using xfs and no problems ever. I've also used btrfs among others and still no issues.
why the fuck u did experiment with that shit when u have good old ext4? xfs is primarily designed for huge capacity storage and other server shit. There is literally no sane reason not to use ext4, and btrfs is still in development.
What is your Distro, op?
I had a lot problems using brtfs and virtual machines on virtual manager.
XFS was made by SGI. You know where SGI is now?
Just use ext4 + LVM. Everything else is a meme. If you want to go pro, get a battery backed controller and SSD and disable cache and barriers.
I've been using ntfs-3g for most of my data for at least 4 years and it got fucked up so that it wasn't even read-only mountable and wasn't fixable by ntfsfix - twice. One year is nothing.
I just want a filesystem with support for compression/deduplication that's not shit.
That would get me back some 300GBs in my hard drive.
Then use zfs.
i dont fucking care who made it, xfs is not for desktop and thats it
This was the kind of filesystem the UNIX-HATERS complained about.
lmfao
...
I had switched from XFS to BTRFS for compression feature. Now I get 12309 every time I do some heavy I/O.
Also I had OP's problem before, and it was easily fixable.
Eh there is a single reasons to use something else see below. Also using ext4 means low level filesystem vulnerabilitiescnnigger backdoors will make you a easier target.
Don't use ext4 on a ssd you moron. It doesn't trim the filesystem properly. Use something like f2fs that is meant for ssd's.
Personally I never tried xfs always thinking it was a meme.
PingFS is the future. True cloud storage. Storing your data on the internet, not on other people's computers.
github.com
...
ZFS is the only filesystem I used for long time periods which NEVER gave an issue.
Love them or hate them, Sun Microsystems tended to build things properly.
Oracle will bury it soon.
Lots of Anons that don't know shit ITT. XFS was the default in IRIX in 1994, long before most of you were alive, and long before multi-terabyte filesystems existed outside of NASA. ext4 wasn't officially stable until 2008 and still needs the occasional bug fix. I have been using XFS on my Linux machines ever since the corruption issues with ext4 around Linux 3.5-3.6 with zero problems to date. Server filesystems are designed to be fast and to not shit the bed. Desktops benefit from this as well.
Just use mount -o discard and a non-meme filesystem, moron.
ZFS is very dependable, but I never liked the amount of bloat that comes with it. It eats way too much RAM, and in my opinion it has no business dealing with RAID, compression, or encryption. I realize that the latter is a minority opinion. Sun was good. Oracle is shit.
Redcuck and cucked derivatives uses XFS by default
NTFS masterrace
...
ReFS cometh
ext4 works great on SSDs and issues trim requests correctly. LVM issues trim requests correctly, too. Swapping it for a special purpose flash-only filesystem designed for phones that is incompatible with everything and not much faster in the best case would be retarded.
You do realize that OpenZFS has very little to do with Oracle, right? Oracle is irrelevant unless you're their customer using their closed-source version.
Will you enlighten us? As I see OpenZFS is not a complete rewrite and essentially the same CDDL-ed codebase as Oracle's ZFS. Oracle has buried Solaris so there's no reason for them to keep developing ZFS. I doubt other sponsors of BSD projects, such as Sony or Apple are interested in this project either.
Yes, I do realize that. To clarify, my comment about Sun and Oracle wasn't meant to be particular to ZFS. I was just poorly expressing my agreement with the user I replied to.
Both, OpenZFS and Oracle's ZFS are based on Sun's CDDL-ed ZFS. Therefore oracle is not relevant. Of course they don't fund public development, that's why others have taken over the development of the open-source version.
Ah, ok. I took it as a point against ZFS. Yeah, Oracle is shit.
To add to that, that's kind of claiming that LibreOffice is shit, because Oracle ruined OpenOffice.
Neither is SGI anymore LOL
Have you tried not having duplicate copies of shit and/or using hardlinks?
On an actually serious note, this is pretty cool: github.com
IRIX is a meme Unix. IRIX is fucking garbage.
Online TRIM is a cool thing you can use if you really hate I/O performance.
Basically pauses your shit every time you delete or update something.
Or you could occasionally bulk-trim (/sbin/fstrim) your FS even though TRIM basically does nothing for performance. (if you disagree: benchmarks or gtfo)
But at least you can run your games at 5x your monitor's refresh rate!
>>>/trash/
Pro Tip: If you can't easily modify partitions with established tools after their creation, it's a garbage FS.
RaiserFS is the only true uncuked FS and he did nothing wrong.
There are a million benchmarks showing it's important. What is wrong with you?
Not the user you responded too.
Well a lot of us don't have trust in benchmarks anymore.
A lot of them are biased because some corporations just want to see they sell increase or because the way of measuring is bollocks.
Why do you even need a benchmark? If you understand the technology you know that wear leveling algorithms will massively decrease write performance and that if the SSD knows when it can skip relocating a block it will be much faster. There's no reason to suspect the cost of keeping the SSD informed would wipe out the benefits since it's just a block bitmap.
>>>Holla Forums
>>>/suicide/
Why don't you "superior OS" fags fight over which of the 1001 filesystems isn't the worst, while we, the NTFS crowd, sit back and enjoy the show from a smooth running machine.
NSA detected.
Please be a troll. Even Microsoft acknowledges NTFS is shit and has tried to replace it. It's the only FS where you need progress bars just to delete 10k files on a SSD.
...
...
lol
The cost of keeping the SSD informed is yuge. It's not just updating a block bitmap, ATA ERASE commands are crazy slow. Turn on trim in ext4 or btrfs and see for yourself.
The cost isn't suspicious, it's noticeable and obvious. The supposed benefit I'm suspicious of, since I've never noticed it.
...
Meanwhile you can use "attic" backup on existing filesystem, mount snapshots.
I use xfs on internal HD no problems, on external usb i have to xfs repair and reboot when it accidentally disconnects, else it won't see the files. Things like kernel panics and remounts not allowed do happen with other FS too.
hello, lesser pc user. Has windows finally gone subscription model, like adobe crap? I dunno, last time I relied on something microsoft it was called applesoft basic.
...
It's not heavy and on good drives it doesn't pause. Don't do it per block via mounting with discards as that's stupid and creates massive numbers of requests, do it periodically with fstrim. Understand how the tech works so you use it correctly.
It seems we're in agreement.
No. The command is extremely fast. Pointlessly creating millions of them for tiny ranges will not be. It's like reading files via mmap() - forcing code to do a context switch per 4k read is going to be very slow despite modern context switches being extremely fast. It's misusing a feature.
Then why does it ruin FS performance compared to bulk trim?
Lots of small reads doesn't destroy SSD performance like issuing erase does. Does erase block everything up like a SMART request?
Bulk trim fucks up performance while it's running.
On a chink SSD without queued TRIM.
...
Damn every time I see this, I can't help but again read it entirely. It's just too good.
Also I can't wait for W10 to finally blow up with some of their future "updates".
XFS is better for large RAID volumes. I have an ext4 root partition on an SSD, and my mdadm /home is XFS.
don't you love it when some kid posts in here, probably owned one PC laptop and thinks because it didn't die in a year its superior?
Although to be fair I have worked with MS faggots sysadmins still spout this shit while their servers are on fire so its essentially just retarded cultism at this point.
it already did multiple times
You have to sit back and watch because you can't do much of anything while defrag.exe is running.
I can't see any reason to use anything outside of ZFS if you can.
Try it on your system regardless and see if the performance is impacted or not, because the multi GB's of RAM people are throwing out there are not actually required for typical desktop use.
I might catch flak for linking this but look at this bullshit even
reddit.com
I'm sure most people here have more than 768MB of RAM and a dual core processor.
I'e got ZFS running on four SATA drives and so far it has been great and its pretty bulletproof.
The only thing is the scrub process can take 10 hours to finish and ZFS recommends weekly scrub on a SATA and monthly on SCSI.
We had a client who had a rather large array with Sun boxes and the scrub process would fuck with backups and other after hours processes when it was running.
Are there any JFS fans here? If so, explain yourself.