[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: nfs re-export issues



Shane Liebling <shane@cryptio.net> writes:

> Hey all, I am working on getting a diskless cluster working and
> everything seems to be fine except for one thing, but first let
> me give you a breakdown of the suspects:
> 
> client - app server - file server
> 
> The client mounts /export/home as /export/home off the app
> server, and the app server mounts /export/home as /export/home
> off the file server.  The client and the app server are running
> 2.2.19 kernels and testing, the file server is a 2.2.18 kernel
> and redhat 6.0.

 Upgrade kernel and nfsutils on the server. 
 
> So rpc.nfsd and rpc.mountd are both being run with the
> --re-export option (which is the only way I can get the client to
> mount them w/o permission denied errors).  My problem is that
> anytime I try to read or write to /export/home on the client, I
> get Stale NFS File Handle errors.

 Funny, I cannot find the --re-export from any manpages. 

 <shameless plug>

 Have you considered SFS ? It can handle re-exports from
 nfsmounts just fine. For more information, see www.fs.net.
 Packages were just installed to non-us/unstable.

 </plug>

> I can read/write just fine on the app server, as long as I am the
> correct user.  All 3 machines have the same /etc/passwd, shadow,
> group except the two servers have all user accounts /bin/false
> for shell, but the client has /bin/bash.
> 
> I have tried rebooting all the machines, start-stopping nfs on
> the servers, tried different kernals on the app server (one w/
> knfsd one w/o).

 What nfs server are you running on the app-server? nfs-kernel-server
 or nfs-server ? If you had nfs-server package installed you were
 running userspace nfs server even if you had the support in
 kernel.

 2.2.19 and nfs-kernel-server with recent nfsutils are known
 good combination. Anything before has bugs more than enough.

        -j



Reply to: