[systemd-devel] consider dropping defrag of journals on btrfs

Lennart Poettering lennart at poettering.net
Mon Feb 8 16:08:24 UTC 2021


On Mo, 08.02.21 10:09, Phillip Susi (phill at thesusis.net) wrote:

> That's a fair point: if btrfs isn't any worse than other filessytems,
> then why is it the only one that gets a defrag?

As answered elsewhere:

1. only btrfs has a cow mode, where fragmentation is through the roof
   for randomly written files

2. only btrfs as a somewhat nice API for this (i.e. a single best
   effort ioctl with no params). (ext4 has a defrag API, but it's
   weird, and xfs I never checked, I never used it)

3. noone was annoyed by journal performance on non-btrfs enough to
   determine if this is worth it.

--
Lennart Poettering, Berlin


More information about the systemd-devel mailing list