Q abt ssh, sshfs, and Stale NFS file handle

[Paul E Condon <pecondon@mesanetworks.net>]

I have debian (lenny and squeeze) running on several hosts that
are interconnected via ethernet. I sit at the console of one
and use ssh and sshfs to control operations on the others.

I am setting up a system to use USB hard drives for backup of
files on all the hosts. I already have working a system for
backup of all hosts onto a single hard drive on a single host.
Now I want to automatically transfer from that permanently
mounted drive onto a USB drive that is mounted using pmount.

So I'm debugging the pmount over ssh. 
I find, after a crash and reboot of the crashed host that I
get an error message when using ls to look at files on the
USB drive via ssh:

cmpq:~# ls -l /media/WDP-5/mystuff/arxiv_5/arxiv/asof/
ls: cannot access /media/WDP-5/mystuff/arxiv_5/arxiv/asof/20091025_010711: Stale NFS file handle
total 4
drwxr-xr-x 2 root root 4096 20100402_083812 00nullbk
d????????? ? ?    ?       ?               ? 20091025_010711

I do not get this message when I view the same directory using sshfs.
The sshfs response to a similar command indicates that there is no
such directory as the one whose name begins with 200...

Restarting the nfs-common on the controlling host does not fix the
problem. Is the problem actually on the controlling host? My
understanding of sshfs is that it is built on top of ssh. How can it
be clearing its hidden cache(s) when ssh does not? I'd rather not have
to reboot the host where I sit.


-- 
Paul E Condon           
pecondon@mesanetworks.net