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

Re: [nfsv4] directory delegations questions/suggestions



On Thu, Mike Eisler wrote:
> 
> 
> --- Spencer Shepler <spencer.shepler at sun.com> wrote:
> 
> > The first one is the error return for GET_DIR_DELEGATION.  As currently
> > specified the operations is to return an error NFS4ERR_DIRDELEG_UNAVAIL
> > if the server is unable/unwilling to provide a delegation.  I would
> > like to suggest that in the case that a delegation is unavailable,
> > that the operation not error out.  Instead, the results struct be 
> > constructed such that one of the cases is a "no delegation" case.
> > This will allow a client to build a compound that can collect
> > directory delegations conditionally without failing the entire
> > COMPOUND.
> 
> makes sense.

I will send out a specific update for this change.

> 
> > The second item has to do with notification of directory entry additions.
> > So along with the new entry's name, there is directory entry location
> > information in the form of cookies.  Thinking through the implementation
> > details of how a server would implement this, it seems too onerous.
> > I realize that the server is not required to implement the notifcation
> > type but it seems as though the decision to suppor this feature should
> > not be how easy it is to implement but more based on the server's 
> 
> ACLs, delegations, and layouts create ease of implementation challenges too.
> As does RPCSEC_GSS and file system migration.
> 
> > workload or general policy setting.
> 
> Some servers might not be able to provide the new entry's directory cookie
> efficiently, though as I pointed out in the con call, even
> conventional UNIX servers know, somewhere, at the time a file is
> created what its directory cookie is. Regardless, it would be 
> reasonable to make the cookie optional.

I will send out a change for this as well to see if it makes sense.

Spencer

_______________________________________________
nfsv4 mailing list
nfsv4 at ietf.org
https://www1.ietf.org/mailman/listinfo/nfsv4