Samba improvements needed

Richard Sharpe rsharpe at richardsharpe.com
Fri Sep 24 00:51:59 GMT 2004


On Thu, 23 Sep 2004, John E. Malmberg wrote:

> Richard Sharpe wrote:
> > On Thu, 23 Sep 2004, John E. Malmberg wrote:
> >
> >>By delaying the open until there is actual data to write or the client has
> >>specified the resulting size, then Method A can be used.
> >
> >
> > [Deleted all the rest of the context because I only wanted to reply to
> > this point.]
> >
> > I do not think you can defer opening the file until there is data to write
> > because of the race that implies. If the file did not exist when the
> > NTCreate&X was processed, that may seem fine, but if a subsequent
> > NTCreate&X comes in with a conflicting share mode, that one may succeed
> > when it shouldn't, especially if it manages to write before you do.
>
> As I understand things now, the share mode conflicts are handled by a
> tdb, so there should not be a conflict.

Yes, that is true ...

> However, at the time of the open request, the vfs module can take out a
> VMS lock on the filespec, and that will provide a cluster wide protection.

... as long as you only run Samba on one node in the cluster ...

> The VFS layer in SAMBA allows much improvement in the VMS port.

Regards
-----
Richard Sharpe, rsharpe[at]richardsharpe.com, rsharpe[at]samba.org,
sharpe[at]ethereal.com, http://www.richardsharpe.com


More information about the samba-vms mailing list