smb.conf issues for 2.2.8

[John E. Malmberg]

In article <CF0913E9C3D53D4E94DE47FDE25C633D048FD44F at hermes.cofiroute.com>,
COLLOT Jean-Yves <jean-yves.collot at cofiroute.fr> writes:
> The template is actually VERY out of date

> With the next release, I'll provide a template that, in addition to be
> up-to-date, without unknown parameters, will reflect the way I
> personally use Samba, i.e. in a "security=DOMAIN" way, and with an
> external = (Microsoft) WINS server.

Multiple template files are probably more useful.  My guess is that the
users are about evenly split between domain users and workgroup users.

My personal desire, and I never had time to act on it what to have a command
procedure that would ask some questions about a new environment, and then
build an SMB.CONF with the correct information.

> The problem is (and I apologize for it) that my main concern is to have
> a satisfactory Samba Server in my own environment, and my boss would not
> be happy if I spent most of my time testing and eventually fixing Samba
> features that we don't even dream of using here (and, as far as I know,
> are not used anywhere).

That is nothing to apoligize for.  With an Open Source project, if someone
wants something different than what is supplied, they have the following
options:

1. Do it them selves.
2. Pay someone else to do it.
3. Beg for someone to do it for free and wait for them.

Your contributions to this project have been significant, and you have
currently become the defacto change control manager of the public distribution.

> When users from the Samba/VMS community encounter problems, I do my
> best to help them, and fix the bugs they encounter. Because nobody ever
> complained about the WINS server feature not working correctly, even if
> I suspect it would not, I infer that nobody uses it, so I forget about it
> for the moment.
> I have definitely no time enough to try everything that can be done
> with Samba.

Which is why more builders and testers need to be recruited.

If you are running a small home network like I do, the WINS function
becomes more important.

Currently with what little spare time I have, I am concentrating on rsync.
and trying to get it to the state where I can build against the same active
build as the rsync developers.  I am currently getting an over 90% build, and
have a client working well enough to keep my local copy of their CVS directory
up to date.

When I started working on SAMBA, I had a network of two PCs and a VAX, and the
PC was the only way I could display X-11 applications.  Now I have two Alphas,
both with DECWindows MOTIF, and do not use the PC much at all any more.

> John E. Malmberg said :
>> It turns out that the hack in past releases NMBD to change the fork()
>> call to return a constant of 0, it appears that might cause NMBD to
>> appear to hang.
>
> That is very possible and most probably true, if you say so. However, I
> have not experienced it, no user I am aware of experienced it either, and I
> don't know what I should do to reproduce that issue. I would need more
> information here.

I am not sure how serious the issue is.  I think that in most cases the
hang induced (if one is induced) is short enough that it does not cause a
protocol time out.  A change to use the TCPIP dispatcher may induce just as
much of a delay as in the present code since the process create time on OpenVMS
is more expensive.  My concern is that I think there may be a possibility of a
deadlock.

I brought up the issue so that others with time can investigate it.

Also slightly related to SAMBA, IIRC, someone on comp.os.vms posted a month
or so ago that they had managed to get a Windows Terminal Server client to
build and run on OpenVMS.

-John
wb8tyw at qsl.net
Personal Opinion Only