Glad to hear it. =o)
> So I've tried to use tsumufs on a samba mount point. It works very
> well and with any changes.
Yeah, we actually implemented the nfsmountcmd and nfsunmountcmd
command line options to make it as generic as possible. In fact, we've
had great success using samba and even sshfs as backend filesystems
without too much trouble. Very interesting stuff. =o)
> Of course ping method should not be same. Do you think there is
> something else which could be depending of our file system type? it
> seems not.
Your line of thinking is correct -- the ping in NFSMount is very
NFS-centric. So when you change the mount and unmount commands to
something different, the ping is wrong.
Ideally, the mount, unmount, and server checks would be implemented in
different classes -- as you have probably seen, we're already doing
all of the NFS specific stuff in one class (NFSMount). To properly
support swapping out the various backends, we should probably get rid
of nfsmountcmd and nfsunmountcmd entirely and replace them with
"backendfilesystem", which would allow for the replacement of NFSMount
with a different class. Once that's done, it would be trivial to
provide different ping methods for each backend type by simply writing
a new class for each type (ie: SMBMount, SSHFSMount, etc.).
--
June Tate-Gans | Don't try to outweird me, three-eyes. I get stranger things
Sysops, SEA | than you free with my breakfast cereal. --Zaphod Beeblebrox