Hi,

Perry is correct.  Softdeps defers i/o, and if deferred i/o fails it has =
to=20
decide what to do about it.  The system call that initiated the i/o opera=
tion=20
has long passed so there is no appropriate mechanism to signal to userlan=
d=20
that something bad has happened.  Also, the filesystem metadata is in a s=
tate=20
of flux, so the code panics.

Just a thought, are those errors at the end of the disk?  Maybe the disk =
label=20
is just a little off.

skd

On Friday 24 January 2003 07:39 pm, Dan wrote:
> Doesn't that sound fun?  Here is a picture of the error:
>
> http://www.fascistadmin.com/misc/panic.jpg
>
> I'm running two mirrored 9.1 gig scsi drives controlled by a Mylex DAC9=
60.=20
> I did a fresh install of 1.6_STABLE (full version at the bottom) and th=
e
> first thing I did was untar pkgsrc.tar.gz and it took forever.  I
> remembered that I hadn't yet turned on softdeps so after enabling them =
I
> rebooted and tried untarring pkgsrc.tar.gz again; that's when I got the
> panic.  The box is still sitting there at the debug prompt if someone n=
eeds
> me to do some operations.
>
> NetBSD 1.6_STABLE (GENERIC) #0: Tue Jan 21 14:04:09 UTC 2003
>
> Dan