Btrfs Restoring Support For Swap Files With Linux 4.21

[pal666]

Personally, with ZFS being as good as it is, I just don't see the point of BTRFS outside of not having to worry about the kernel/module upgrade annoyances ZFS brings...I'd rather deal with those annoyances than a not-as-good file system.

but btrfs is better than zfs. zfs is obsolete even if it were intree. i have no use for fs which is unable to change its size

[pal666]

A more recent article pointed out the corruption was caused by BLK-MQ, outside the ext4 driver

ok, that makes ext4 users who lost their data much happier

[pal666]

I am a moron. I rather not deal with swap partitions. I don't want any extra partitions. I rather just not even think about swap.
The last thing I wanna do is trying to figure out how many gigabytes my swap partition should be, if it should be before or after my system partition, if it matters if its a HDD or SSD, etc.

well, you shouldn't even bring swap subject to discussion then. installer created partitions for you, including swap

[pal666]

IMHO, the benefits of ZFS outweigh the negative aspects. Besides, ZFS on Linux should only get better and gain more performance as they creep towards the 1.0 release.

you've listed zero benefits (there is nothing in your list which is unavailable on btrfs, except you'll have to use full disk/partition encryption for now)
and btrfs is also only getting better and gaining more performance

[pal666]

I do use it on my Linux servers and it works VERY well. So far my experience has been way better than btrfs.

there are enterprise linux distros which default to btrfs, so you can leave your anecdotal evidence under the blanket

[pal666]

It wasn't ext4 fault.

not having redundancy and checksums was ext4 fault

[pal666]

Its development is still way faster than btrfs'.

its development has zero speed, as can be seen in kernel commit log

[pal666]

The ZFS wiki recommends using NOOP

so instead of worrying about schedulers you've chosen to worry about wikis of some out of tree unsupported crap?

[ultimA]

pal666 , some false facts and half-truths from you.

there are enterprise linux distros which default to btrfs, so you can leave your anecdotal evidence under the blanket

No, there is actually only a single one, Suse. There used to be Red Hat too, but they removed support for btrfs after "trying it out" and seeing it still has a lot of issues. And even on Suse, while btrfs is the default for the root volume, they recommend MD RAID instead of BTRFS for raid configurations. So much for enterprise-readiness.

its development has zero speed, as can be seen in kernel commit log

Of course, because ZFS is being developed outside of the kernel tree. So the kernel commit log has little bearing on the matter. ZFS-on-Linux is being actively maintained and developed, if you know where to look.

[edgmnt]

Has anyone benchmarked regular swapfile vs. swap on loopback-mounted file? I figure filesystem compression might offset some of the overhead associated with loopback, that'd be interesting to see.