cifsclient

[Denise Spinger]

With a CIFS Windows share mounted to a unix directory (/sbcdf213/natu$), no files appear. The following messages occur when reporting on this mount from a bdf(1M) command:
NFS getattr failed for server sbcdf213: RPC: Timed out
NFS fsstat failed for server sbcdf213: RPC: Timed out
bdf: /sbcdf213/natu$: Connection timed out

A ping(1M) to the server is successful and the IP address is correct for the Windows server, and this share can be accessed Windows pc client.
Also, this connection was working ok recently. If i reboot the server i can get it work (i did it the last time it hapenned), but i can not reboot it now. The following attempts
were made to recover but were not successful:

# cifsclient stop
Shutting down CIFS Client daemon:
unmounting...
done.
# cifsclient force_umount /sbcdf213/natu$
nfs umount: nfs_unmount: /sbcdf213/natu$: is busy

# cifsclient start

# umount -aF cifs
umount: FStype does not match any /etc/mnttab entries

# cifslogout
Logging out User: Server not connected!

# cifsmount //sbcdf213/natu$ /sbcdf213/natu$ -U xxxxxx -P xxxxxx

NFS getattr failed for server sbcdf213: RPC: Timed out
Mounting Share: Device busy
Connecting Share: DOS: Share not exported by server



How can this CIFS connection be restored?

[doug mielke]

it may be that the only way to get rid of the busy message is to reboot. However, if you can tolorate it, you can mount the share to another mount point. (cifs fails for us so often we have our apps looking for the shares in 2 locations, in case we get this problem.

[Denise Spinger]

Thanks for the suggestion, but we need to use the same mount point, because of the application which uses it. It´s an ETL application and i do not know enought about it, but the user who needs it told me that it would get a lot of work to change the mount poin

[Steven E. Protter]

You have to boot to clear this, more than likely.

You might be able to identify and kill a Unix process as follows:

fuser -cu /mount_point

use the -k option to kill all processes. Better to issue individual kill commands until the contention is cleared.

Take away th c option if its not amounted file system.

Be careful, you can do serious damage with that command.

SEP

[Phillippe J. Welsh]

I just had the same problem, but just figured out how to correct it WITHOUT a reboot (for me). As root:

1. "cifsclient stop"
2. "cifsclient force_umount YOUR_MOUNT_POINT"

and my problem (and phantom mount point) went away. Found this handy little option by reading the info from "cifsclient -v". I should warn that the help states "...this is an emergency procedure to be used only in case of failure of the standard umount commands...".