Tux3 File-System Gains Initial FSCK Implementation

[gilboa]

LVM snapshots do not guarantee filesystem integrity. There is a filesystem freeze hack in Linux that is used to try to prevent filesystem corruption, but it does nothing it you are using a hypervisor and then snapshot from the host.

True, but at least in reality, one can run fsck before calling snapshot or even while the snapshot is active.
Granted, ZFS does have the upper hand here (and maybe, in the long run btrfs), but as it stands, LVM+ext4 is a *very* powerful (and proven) combination.

- Gilboa

[ryao]

It is quite possible that Tux3 will get to incremental and online fsck before
Ext4 does. (There you go, Ted, that is a challenge.) There is no question that
this is something that every viable, modern filesystem must do, and no,
scrubbing does not cut the mustard. We need to be able to detect errors on the
filesystem, perhaps due to blocks going bad, or heaven forbid, bugs, then
report them to the user and *fix* them on command without taking the volume
offline. If that seems hard, it is. But it simply has to be done.

ZFS's online scrub is sufficient because of it is atomic copy-on-write transaction commit, uberblock history and ditto blocks.

Concerning missing snapshots, ext4 doesn't really it need it if you're using LVM.
(As I type this, the LVM on this workstation is busy merging back a snapshot take before a botched upgrade attempt from F17 to F18...)

- Gilboa

LVM snapshots do not guarantee filesystem integrity. There is a filesystem freeze hack in Linux that is used to try to prevent filesystem corruption, but it does nothing it you are using a hypervisor and then snapshot from the host.

[gilboa]

This file system have a lot of potentiality i hope that everything work the best until, finally, will be ready.

I believe that would have the potential to become an enterprise file system something that is not ext4 (since important features such as snapshots are not present) but remained far lighter (so usable even on old machine with few resource) respect Btrfs (that is a sort of an elephant so we need not wonder that despite the support it receives isn't never ready as default file system and anyway require too much resource).

Concerning missing snapshots, ext4 doesn't really it need it if you're using LVM.
(As I type this, the LVM on this workstation is busy merging back a snapshot take before a botched upgrade attempt from F17 to F18...)

- Gilboa

[gilboa]

Umm. You cannot avoid errors if say your hard disk is failing.

... let alone random bit errors due to, say, background radiation.

- Gilboa

[alelinuxbsd]

This file system have a lot of potentiality i hope that everything work the best until, finally, will be ready.

I believe that would have the potential to become an enterprise file system something that is not ext4 (since important features such as snapshots are not present) but remained far lighter (so usable even on old machine with few resource) respect Btrfs (that is a sort of an elephant so we need not wonder that despite the support it receives isn't never ready as default file system and anyway require too much resource).

[Ericg]

I mean, atomic operations and similar things to avoid errors which are not hardware-sided.

Oh =P Still, bugs happen. Better to have a working fsck early on and expand upon it as you add more and more features haha

[Cthulhux]

I mean, atomic operations and similar things to avoid errors which are not hardware-sided.

[Ericg]

Online fsck is nice. Good file systems avoid errors though.

Errors happen, bugs happen. Its the nature of software. You do the best you can, but a bug will always find a way in. Plus hardware fails as Rahul pointed out. Thats why its not a "We have the perfect filesystem. Zero errors!" approach. Its a "This is reality. Shit happens. We need to protect against the shit." approach

[RahulSundaram]

Online fsck is nice. Good file systems avoid errors though.

Umm. You cannot avoid errors if say your hard disk is failing.

[Cthulhux]

Online fsck is nice. Good file systems avoid errors though.