Le 24/08/2017 à 09:46, Narcis Garcia a écrit :
El 24/08/17 a les 01:03, Adam Borowski ha escrit:
Hi!
I'd like to recommend another improvement: let's make the installer default
to noatime for fstab it creates.
In the past, atime updates used to ruin performance. Thanks to work by Ted
Ts'o and others, that penalty has been greatly reduced (but not eliminated)
by two options:
* relatime (on by default): atime is not updated unless atime<=mtime or
atime<NOW-24h
* lazytime (off by default): atime and mtime updates are postponed by up to
30 minutes, unless there's memory pressure or the inode has to be written
for another reason
But alas, there are more considerations than just performance:
* atime is really nasty for any CoW filesystems (btrfs, zfs, lvm snapshots,
qcow2 snapshots, deduped thin storage, etc). For a /usr-y mix of files,
it costs around 5% of disk space per snapshot. relatime offers no help as
its threshold (24h) just happens to match the most popular snapshot and
cronjob frequency (once per day). Even worse, it can make _reading_
things result in ENOSPC!
* atime murders media with sharply limited write endurance, such as SD
cards. For this reason, every SoC pre-made image I've seen ships with
noatime.
And what do we get in return?
* the "vote" field in popcon. This somewhat benefits the distribution (lets
us spot low-use packages installed by default, such as xterm (53.60% inst,
6.67% vote) or tcpd (95.20% inst, 9.85% vote)) but gives no gain to
individual users.
* some rare forensic uses
* new mail notification on login
The last thing has recentlish (POSIX-2008) gained a solution: mail readers
can be patched to manually calls futimens(f, {UTIME_NOW, UTIME_OMIT});
stretch/ascii already sports mutt patched this way -- not sure what other
mail clients are likely to be used by local users thus worthy of patching.
Sample patch:
https://github.com/neomutt/neomutt/commit/816095bfdb72caafd8845e8fb28cbc8c6afc114f
Because of stretch's freeze, I did not manage to write+push upstream patches
for any client other than mutt, I guess it's time to resume.
Obviously, an admin who thinks he actually has an use for atime is free to
edit fstab.
So, what would you folks say about defaulting to noatime?
Meow!
Since 10 years ago I'm setting "noatime" to all mounts in thousands of
GNU/Linux installations.
Yes, please set "noatime" as default for any volume.
+1
When I first formatted all partitions with Reiserfs, which was the
first journaled filesystem, it was recommended to set noatime for
performance. There was no visible drawback. With SSDs, atime is mostly
harmfull. I would recommend also nodiratime, which is not implied by
noatime. And I think that none of these are implied by mounting rdonly.
Didier
_______________________________________________
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
