Check this out:
df -h
Filesystem size used avail capacity Mounted on
/dev/dsk/emcpower1a 49G 16384E 79G 36105706404683% /u02

I wanted to restore my oracle databases from tape so I first removed
all of the database files in /u01, /u02, and /u03. I ran the restore
and recovered all the files but ended up with a file system full on
/u02:
/dev/md/dsk/d30 15G 2.5G 12G 17% /u01
/dev/dsk/emcpower1a 49G 49G 0K 100% /u02
/dev/dsk/emcpower0a 15G 98M 15G 1% /u03
That's when I remembered that I'd forgot to stop oracle first so the
oracle processes were still holding the old database files open and
therefore they hadn't been deleted.

I stopped oracle and then checked df:
/dev/md/dsk/d30 15G 2.5G 12G 17% /u01
/dev/dsk/emcpower1a 49G 16384E 79G 36105706404683% /u02
/dev/dsk/emcpower0a 15G 16M 15G 1% /u03
All I can say is WOW!

I umounted /u02 which took a very long time - pressing ^C didn't return
a command prompt, nor did ^\. Eventually /u02 did unmount.

I ran a fsck /u02 which found no problems. Remounted /u02 and
everything is fine again.

I've seen this problem once before on this same box when I first
switched to Solaris 10.

For what it's worth, this is the output from mount:
/u01 on /dev/md/dsk/d30
read/write/setuid/devices/intr/largefiles/logging/xattr/onerror=panic/dev=154001e
on Fri Mar 3 12:01:02 2006
/u03 on /dev/dsk/emcpower0a
read/write/setuid/devices/intr/largefiles/logging/xattr/onerror=panic/dev=2680000
on Fri Mar 3 13:29:01 2006
/u02 on /dev/dsk/emcpower1a
read/write/setuid/devices/intr/largefiles/logging/xattr/onerror=panic/dev=2680008
on Thu Mar 23 13:28:41 2006

Any ideas?

On Thu, 23 Mar 2006 03:59:12 +0100, Roland Gerlach
<roland.gerlach@gmail.com> wrote:

> Check this out:
> df -h
> Filesystem size used avail capacity Mounted on
> /dev/dsk/emcpower1a 49G 16384E 79G 36105706404683% /u02
>
> I wanted to restore my oracle databases from tape so I first removed
> all of the database files in /u01, /u02, and /u03. I ran the restore
> and recovered all the files but ended up with a file system full on
> /u02:
> /dev/md/dsk/d30 15G 2.5G 12G 17% /u01
> /dev/dsk/emcpower1a 49G 49G 0K 100% /u02
> /dev/dsk/emcpower0a 15G 98M 15G 1% /u03
> That's when I remembered that I'd forgot to stop oracle first so the
> oracle processes were still holding the old database files open and
> therefore they hadn't been deleted.
>
> I stopped oracle and then checked df:
> /dev/md/dsk/d30 15G 2.5G 12G 17% /u01
> /dev/dsk/emcpower1a 49G 16384E 79G 36105706404683% /u02
> /dev/dsk/emcpower0a 15G 16M 15G 1% /u03
> All I can say is WOW!
>
> I umounted /u02 which took a very long time - pressing ^C didn't return
> a command prompt, nor did ^\. Eventually /u02 did unmount.
>
> I ran a fsck /u02 which found no problems. Remounted /u02 and
> everything is fine again.
>
> I've seen this problem once before on this same box when I first
> switched to Solaris 10.
>
> For what it's worth, this is the output from mount:
> /u01 on /dev/md/dsk/d30
> read/write/setuid/devices/intr/largefiles/logging/xattr/onerror=panic/dev=154001e
> on Fri Mar 3 12:01:02 2006
> /u03 on /dev/dsk/emcpower0a
> read/write/setuid/devices/intr/largefiles/logging/xattr/onerror=panic/dev=2680000
> on Fri Mar 3 13:29:01 2006
> /u02 on /dev/dsk/emcpower1a
> read/write/setuid/devices/intr/largefiles/logging/xattr/onerror=panic/dev=2680008
> on Thu Mar 23 13:28:41 2006
>
> Any ideas?

this is bug:

6362734 df output corrupt after under heavy stress

a fix is under the works.

---
frankB

Thanks for the reply. When I seached sunsolve for 6362734, I also
found 6391678 df miscalculates used and capacity values during large
deletes which claims that "Depending upon the amount of data being
deleted, df can see this problem for around 30 seconds before it
eventually reports the statistics correctly". So I didn't need to
umount the filesystem after all.

Thanks once again.