On Fri, Jul 01, 2005 at 05:16:59PM +0200, Nicolas Joly wrote:
> 
> I noticed that df(1) can show diffrent values for a NFS filesystem if
> mounted with v2 or v3 protocol :
> 
> Filesystem                                     Size      Used     Avail Capacity  Mounted on
> xx.xx.pasteur.fr:/vol/banques/.snapshot/prod   596G     663G      -67G   111%    /mount/banks2
> xx.xx.pasteur.fr:/vol/banques/.snapshot/prod   596G     -15T       16T1475378468640%    /mount/banks3
> 
> Our main storage device is a NetApp filer R150 with about 12TB; it
> holds the biological databanks in a 2.3TB volume (= partition). We
> export a read-only snapshot from that volume to all our working
> nodes. This method, allow the databanks maintainers to update/work on
> the datas without disturbing the users ...
> 
> Sometime, the modifications are really big and the reserved space for
> the snapshot is hit; then, the filer automatically extends the
> snapshot size by taking needed blocks in the underlying volume.
> 
> In that case, i noticed that the NFSv3 mounted partition give
> erroneous values; but the same partition, NFSv2 mounted, has the
> correct values.

After some more experiments, it seems that the NetApp filer is sending
bogus values for available size with NFSv3. Will try to bug NetApp ...

-- 
Nicolas Joly

Biological Software and Databanks.
Institut Pasteur, Paris.