Re: Re: File system defragmentation. Something that can resume after stopping.

Michael <[email protected]> Thu, 14 May 2026 18:30:51 +0100

Newsgroups	gmane.linux.gentoo.user
Message-ID	<2413994.ElGaqSPkdT@rogueboard>

On Thursday, 14 May 2026 17:18:39 British Summer Time Holger Hoffstätte 
wrote:
> On 2026-05-14 12:30, Michael wrote:
> > There's this archived paper from a 2007 Linux Symposium which
> > explains how linux filesystems intelligently delay writing data to
> > disk until the last moment so that data is allocated as
> > contiguously as possible.  Also, after a
> That (delalloc) is unfortunately not even half the story. XFS is still
> the only filesystem (on Linux) that does persistent preallocation on
> regular writes, and splitting the filesystem into multiple AGs
> (allocation groups) is much more effective for *long-term* mixed-size
> free-space management than whatever ext4 does even today.
> 
> So while we're posting papers:
> https://www.usenix.org/conference/hotstorage19/presentation/conway
> 
> Enjoy :)
> 
> -h

Thank you for this!  I was surprised to read BTRFS was tested to be 
better than ext4 and ZFS on HDD and better than ext4, XFS and F2FS on an 
SSD.  O_O

Does anyone know if kernel fscrypt has been added to btrfs following 
this work?

https://lore.kernel.org/linux-btrfs/
20240124195831.GA1212739@perftesting/T/

I don't see this included in CONFIG_BTRFS_EXPERIMENTAL on the stable 
kernel.

signature.asc (application/pgp-signature, 870 B)