Subject: Re: UVM/other problems for desktop users in current?
To: None <current-users@netbsd.org>
From: Dan Melomedman <dan%dan.dan@devonit.com>
List: current-users
Date: 12/19/2002 15:53:43
Frederick Bruckman wrote:
> That's the ideal. With the write cache on the disk enabled, it doesn't
> quite work that way, but then, it seems to be no worse than the
> non-softdeps case. With a write back cache on the drive (or
> controller), even mounting the file system "sync" isn't safe, in that
> regard.

RAID controllers with battery-backed write-back cache are perfectly safe
as long as the controller itself is not responsible for the crash. When
the machine is brought back, the cache will be flushed back to the disks.

Write-back cache on the controller without a battery backup
is obviously very risky. Obviously, write-back cache on disks
should be disabled unless the disk holds unimportant data.
A good RAID controller will let you turn this on or off.

> The performance boost is a side-effect, that comes from no longer
> writing the metadata synchronously. NetBSD, by default, still writes
> the metadata synchronously and the filedata asynchronously. I gather
> the main people recommend, or use, softdeps on NetBSD is to get
> roughly equivalent performance to mounting the file systems "async",
> but without throwing all caution to the winds.
> 
> Frederick
>