Move var/locks/.nmbd to var/nmbd for 3.6.0?

simo idra at samba.org
Tue Jun 14 11:53:33 MDT 2011


On Tue, 2011-06-14 at 14:06 +0200, Volker Lendecke wrote:
> On Tue, Jun 14, 2011 at 07:59:10AM -0400, simo wrote:
> > On Tue, 2011-06-14 at 07:46 +0200, Volker Lendecke wrote:
> > > On Tue, Jun 14, 2011 at 03:31:23PM +1000, Andrew Bartlett wrote:
> > > > Additionally I've also added a patch to move the PREFIX/var/locks/.nmbd
> > > > socket directory to PREFIX/var/nmbd
> > > 
> > > Why this change? We have /tmp/.winbindd, which I modeled
> > > this after. Can you give me a reason what functionality a
> > > name ".nmbd" breaks? And, by the way, this name went in
> > > months ago, and it seems to work fine.
> > 
> > FWIW /tmp/.winbindd is problematic and is going to hurt Samba as soon as
> > distributions will start using named spaces for /tmp so each user has a
> > different, personal directory.
> 
> Wait -- /tmp per user? I would guess that Samba is not the
> only program that will severely suffer here. If everybody
> gets his own private directory tree, I think we must make
> winbind listen on TCP. But then you could also add network
> namespaces, multiple 127.0.0.1.

I don't get why, /tmp is not the only directory in a filesystem.
As I said in Fedora/RHEL we simply moved the pipe to /var/run/winbindd
what's the problem with that ?

> Simo, what is the standard way these days to offer services
> on a Linux system? DBUS?

DBUS is a unix pipe like we have in winbindd just with a different more
generic protocol. You are welcome to use it if you like but it is not
necessary.
You may mean the system message bus (that uses libdbus as the
transport), but that's not meant to carry a lot of data, so it is
inadvisable to sue it for the winbindd protocol.

Simo.

-- 
Simo Sorce
Samba Team GPL Compliance Officer <simo at samba.org>
Principal Software Engineer at Red Hat, Inc. <simo at redhat.com>



More information about the samba-technical mailing list