I’ve been using computers since the late 80’s. Since like early '89. Literally since I was around 7. And I’ve seen file system’s come and go, some lasting FAR LONGER than they were probably ever intended. (I’m looking at you FAT)
For all the operating systems or at least a good majority of the main ones, what do you think is the most stable?
Just wanting to start a conversation. I’m genuinely interested in people’s opinions. And somewhere I’ll get to thinking of my own and post mine.
Ext2/3/4
Xfs
Jfs2
I’d vote for Xfs.
Zfs is a good reliable file system. Requires some reading and has it’s caveats.
ZFS solves a lot of problems the others don’t.
The memory requirements aren’t that bad unless you try and do live de-dupe which is normally not advisable.
The rest aren’t even close.
BTRFS is almost as flexible as ZFS, and may have more compatibility across the Linuxsphere.
The new MS ReFS is supposed to be good, but in my small experience with it on ‘10 pro, has a few bugs, so I switched back to straight old NTFS on my Win partition
Say what you want about windows but ntfs is pretty stable. Multiple copies of the file table mean that retrieving data from fucked drives is usually possible. I’ve never had the partition go bad without the drive also failing. In fact there’s usually no need to format your windows drives. Want to reinstall? Just install right over the old one.
Of course Linux partitions have proven to be more useful in feature set. I won’t argue that ext is probably just as stable (if not more) than ntfs. I’m not so sure about the relative newcomers like zfs and btrfs but I’m excited to see them mature into something great. They’re already so much more useful than the usual staples and are stable enough right now that they are totally useable without too much worry.
NTFS.
Ext4 pretty good too.
Haven’t had the best of luck with FAT.
I’m still trying to find a reference for this, so someone can help me confirm or deny this postulate…
What I must first say is that I believe in terms of I/O, you cannot go wrong with any Journalling filesystem. The time it would take you to check the disk is much lower than boot time fscheckers.
Each journalling file system is going to have pros and cons, so I would research them and get familiar with them. Ext3 is a very popular one, so there’s plenty of white paper and specs and typically is backward compatible, although in terms of performance it’s not as fast as XFS. But from your post I surmise that you don’t want a Formula-1 race-car, you want the nearly indestructible Honda.
Postulate: I have read somewhere that either the Windows filesystem or specific windows programs sprawl files all over the disk, resulting in the R/W heads needing to seek everywhere for specific information, which will wear out your disk faster. However, I would need a citation to put forward this postulate in full confidence.
Defragging will minimize this, assuming you’re on spinning rust.
ext3/4
is nice plenty of tuning and options. Plenty of disaster recovery guides have been creates for those.
NTFS/ReFS
ReFS is the new NTFS, but too little info on recovery available.
BtRFS
One of the most feature complete partition, but lacks user experience.
beyond those you may want to look at proprietary partitions as they are likely to be much better.
You can find comparison of feature levels of each partition here:
As I am reading this thread through I realize how little I know about these
If you guys come to some agreements over these, think that would be great wiki thread, especially if there is some exotic new thing for SSD’s 
Dunno if i’d be trusting BTRFS
https://forum.rockstor.com/t/zfs-filesystem-for-datastorage-due-to-btrfs-raid56-bug/2095
Sure, 2016, but raid5/6 being broken only 2 years ago whilst ZFS has been enterprise deployed since about 10-15 years ago…
edit:
first link i googled for BTRFS broken. Because i remember a while back significant problems in BTRFS with features that were supposedly feature comparable to ZFS depending on who’s BS you believe.
more relevant info
2c. but over the years i’ve heard/read about WAY too many bugs with BTRFS to trust it. those who do not understand ZFS are condemned to reinvent it poorly, etc…
IFAIK raid5 or 6 on btrfs was never advertised as working. I’ve had good experience running raid1 on btrfs. Very easy to expand the array too.
FWIW I’ve never really had trouble with NTFS after version 5 came with (I think) Windows 7.
Wendell did a few videos on BTRFS back in the Tek Syndicate days. I was convinced then it wasn’t ready yet. I’d stick with a format that’s been proven, in this case, ZFS, until BTRFS can show it’s as reliable.
ZFS, no contest.
btrfs loses data
bcachefs is still an open research project
reiser is garbage
ntfs, fat are ancient and not feature complete
xfs is incompatible with fsck and fragile
ext, hfs, etc is fine but has no built in Cow or redundancy features
ReFS isn’t actually journaling or CoW, it just has metadata that it does nothing with. It’s just an excuse to break compatibility with OSX, Linux, and other operating systems.
have fun with balances, random depop and incorrect lz4 compression reporting, and a bunch of broken, band-aided, then rebroken features
if you dislike Zfs for whatever reason, mdadm + ext4 is far more reliable. Hell, It’s even used in production by responsible adults.
Btrfs: It stops working so you don’t have to
Btrfs: memories shouldn’t be cheapened with records
Btrfs: Live in the moment, and only in the moment
Btrfs: metadata is just too ‘meta’
Sorry kids we had to put him down.
It would be nice for Windows to get an up to date filesystem. But I guess there is no real reason for it to per Microsoft
Anyone wdo needs a serious os to do work isn’t on windowsin the first place. Thats why its just the gamer os.