[lsb-discuss] statfs specification addition:review

Kevin Dankwardt k at kcomputing.com
Thu Dec 21 12:10:32 PST 2006

>semantics like 'is this a network file system' or 'is this a cluster
>file system' might be more general and less-well defined since they
>actually represent attributes of the backing datastore for the file
>system.  However even these can be broken down into concrete terms like
>'reliability', 'performance' (network file systems generally not be

I also say that some of these attributes are features of the backing store. For example, what if EXT3 is put on a network block device. It seems that some of these attributes need to percolate down to the device upon which the file system instance lives. The attribute of "networkness", or even "file lock coherency" may not necessarily be a question for the file system itself.

Similarly, a virtual machine whose entire image is actually being provided by a file server may not truly have any of the attributes such as O_DIRECT, ability to write, networkness, file lock coherency, etc. as valid no matter what the file system itself thinks.

At the least, it seems to me, that there may have to be some disclaimer around querying of these attributes - 'these attributes only apply when the underlying devices follows the semantics ...".
-kevin dankwardt

More information about the lsb-discuss mailing list