Question about Cyrus, ext3, and Linux kernel 2.4.18 bug

[James Ralston]

On 2003-04-16 at 22:38:15-0400 Lawrence Greenfield <leg+ at andrew.cmu.edu> wrote:

> Sigh.  This is a well known problem with ext2.  It is impossible for
> any application to maintain sane consistency guarantees with ext2 in
> the default async mode.  (It's not clear mounting ext2 in "sync"
> mode is sufficient for any serious application.)

It's usually a big performance hit, yes.

> All the Linux journalling filesystems implement softupdates-like
> semantics and do not require an fsync() on a directory to force
> filenames to disk.  There are various reasons why doing so is
> unreasonable, which is why MTAs and Cyrus refuse to do so.

Could you enumerate the reasons?

> I can't think of any sane reason why ext3 authors would want to
> weaken the semantics they export to applications now.

Don't count on it.  There's a fair amount of precedence in Linux that
previous behavior will be cheerfully tossed out the window if it
results in what Linus et. al. think is a better system.  Unless
there's a relevant standard which asserts that fsync() must implement
softupdate-like behavior (I know of none), it's open to change.

James