Error unmounting file system

[Raj Mithal_1]

I have tried removing mnttab file but it regenerates with the entry I don't want.
I assume it comes from the kernal. Is there a way to flush this.
I know a re-boot will solve this but there is too much at stake to call for a re-boot.
SAM & Veritas ADm gui give me the Block Device error.
The file system is empty & I want to get rid of it so I can put a new volume on this DG.
Can anyone see a problem in the following:-

hcp1 / # umount /u/oradata/unel/arch
umount: cannot unmount /u/oradata/unel/arch : Block device required
umount: return error 1.
hcp1 / # umount /dev/vx/dsk/uneldg/unelarchvol01
umount: cannot unmount /u/oradata/unel/arch : Block device required
umount: return error 1.
hcp1 / # hcp1 / # umount /dev/vx/rdsk/uneldg/unelarchvol01
umount: cannot find /dev/vx/rdsk/uneldg/unelarchvol01 in /etc/mnttab
cannot unmount /dev/vx/rdsk/uneldg/unelarchvol01
hcp1 / # vxumount -F vxfs -o force /dev/vx/dsk/uneldg/unelarchvol01
vxfs vxumount: cannot set force unmount for /u/oradata/unel/arch: Invalid argument
hcp1 / # vxumount -F vxfs /dev/vx/dsk/uneldg/unelarchvol01
umount: cannot unmount /u/oradata/unel/arch : Block device required
umount: return error 1.
hcp1 / # grep unel /etc/mnttab
/dev/vx/dsk/uneldg/unelarchvol01 /u/oradata/unel/arch vxfs ioerror=mwdisable,qio,delaylog 0 1 1164621836
hcp1 / # bdf|grep unel
/dev/vx/dsk/uneldg/unelarchvol01
32768000 11768842 19699220 37% /u/oradata/unel/arch
hcp1 / # ll /dev/vx/dsk/uneldg/
total 0
brw------- 1 root sys 2 0x45ee59 Nov 29 11:21 unelarchvol01
hcp1 / # ll /dev/vx/rdsk/uneldg/
total 0
crw------- 1 root sys 61 0x45ee59 Nov 29 12:24 unelarchvol01
hcp1 / # fsck /dev/vx/dsk/uneldg/unelarchvol01
fsck: /etc/default/fs is used for determining the file system type
file system is clean - log replay is not required
hcp1 / # fsck -F vxfs -m /dev/vx/rdsk/uneldg/unelarchvol01
vxfs fsck: sanity check: /dev/vx/rdsk/uneldg/unelarchvol01 OK

[Steven E. Protter]

Shalom,

1) Something or daemon or process is keeping the mount point active so it will not umount.

2) Don't run fsck on a filesystem that is mounted, this can destry it.

Best bet: modify /etc/fstab boot the box then do the work.

SEP

[Michael Steele_2]

Use this command to verify and kill any processes still attached to a mounted filesystem.

# fuser -cu /mount_point
# kill -9 pid

Regarding your /etc/mnttab file being out of sync with your /etc/fstab file or bdf report, rename /etc/mnttab and make a new one this way:

# mv /etc/mnttab /etc/mnttab.old
# mount -a

Note: Don't edit /etc/mnttab.

[Raj Mithal_1]

I know a re-boot will solve the problem but I can't do it till the next maintenance slot unless it's an absolute emergency of which this is not.

I have deleted /etc/mnntab and mount -a recreates it with the unwanted entry.

I think that the file is open but I can't see by what or where. The below comes back empty:-
hcp1 / # fuser -cu /u/oradata/unel/arch
/u/oradata/unel/arch:

hcp1 / # lsof /u/oradata/unel/arch
hcp1 / #

I think the solution lies in one of:-
- Force close the file and destroys
- Create a new Block Device (not sure where or how)
- Clear the mount point from the kernel.

[Chan 007]

Raj,

Just check for any oracle(archive) rogue process running on the system, try top or ps -ef |grep arc.

That may be doing all these junk work. If you have a cluster try to move the package to other system and see if you can unmount it.

If not the only option is reboot.

Chan

[Raj Mithal_1]

No rogue process found

[Yogeeraj_1]

hi,

can you try to rename it before retrying?

mv /u/oradata/unel/arch /u/oradata/unel/arch2

hope this helps!

kind regards
yogeeraj