On Sun, 12 Aug 2001 12:49:59 +0200 Bram Moolenaar <Bram Moolenaar
> The "pulling the plug" test is a very good way to check what would happen if
> the power fails unexpectedly. E.g., when a fuse blows.
> > 5. The delayed write you talk about is the norm for ALL filesystems
> > operating on spinning disks. If you don't delay writes in a filesystem
> > then you will be here until Christmas responding to this email.
> > Now XFS has delayed allocation which is different.
> More technical reasons, which explain why it works this way, but give no
> reason to want it that way.
If you feel this way then I suggest you stay with MSDOS which always
uses synchronous writes and suffers performance accordingly.
Every multi-user system I've ever come across works the way you are
objecting to - they have to to get adequate performance, and it does
not matter whether the wait 100 seconds or 10 mSec - there is still a
chance of failure between the OS saying 'OK, I've done' and the data
actually being written to disc.
The only practical way to achieve what you want in a multi-user OS is
to use transaction mechanisms as used in database systems, and that
would require a redesign of all relevant applications because they
would have to pass 'start of transaction' and 'end of transaction'
messages to the OS.
Keith Matthews Spam trap - my real account at this
node is keith_m
Frequentous Consultants - Linux Services,
Oracle development & database administration