My proposed path to a Beta for Samba 4.0

Andrew Bartlett abartlet at samba.org
Fri May 11 03:27:48 MDT 2012


On Fri, 2012-05-11 at 11:38 +0300, Alexander Bokovoy wrote:
> On Fri, May 11, 2012 at 10:56 AM, Andrew Bartlett <abartlet at samba.org> wrote:
> > As I have mentioned before, we are very close to making a beta release
> > of Samba 4.0.
> >
> > We haven't done betas of Samba for quite some time (the 3.x series
> > hasn't used alpha or beta releases since 3.0), so I should be clear what
> > I mean by a beta, as I seek concurrence or comment.
> >
> > My proposal is that the release of Samba 4.0 beta 1 will be much like
> > the alpha and TP series before it, but the transition marks the maturity
> > of the long-awaited integration of the smbd file server, currently being
> > developed under the s3fs project.
> Agree.
> 
> > Before we can make a beta 1 release, we need to merge the patches I
> > proposed and make the use of s3fs the default mode of operation for a
> > new provision.  To avoid disruption to existing production users, I
> > propose that this be done by writing entries in the smb.conf at
> > provision time, rather than changing the defaults inside Samba.
> Agree.
> 
> > Once these two steps are made, I think we should make a release.  I do
> > not propose to wait on any other features/patches/bugs, but to instead
> > focus on making a beta2 as soon any any other significant changes are
> > ready.
> I've got MIT Kerberos build working during SambaXP. We have few
> patches pending in Simo's tree waf-mit-work (and one more in my local
> tree due to flaky networking at Freizeit Inn). Out of these patches
> about 4-5 needs to be cleaned up and then whole batch can be
> submitted. This would take a week or so, not more.
> 
> Heimdal build keeps being the same so no issues on that.

Great, I'll take a look at it.  Please check with Jelmer that the system
Heimdal build keeps working.

> One particular issue that we see in git master now is setgroups()
> crash which autobuild didn't catch due to being run under
> non-privileged account. The bug affects passdb modules based on
> ldapsam code (ldapsam itself and FreeIPA's ipasam at least) so fixing
> it for ldapsam is important for s3fs combined use.

s3fs won't hit ldapsam issues (the AD DC uses only
auth_samba4/pdb_samba4), but regardless, this looks like a nasty bug. 

> http://abbra.fedorapeople.org/.paste/smbd.panic.log is what I'm
> getting with MIT build and ipasam in use (Guenther saw the same crash
> in Heimdal build in master). This will be worked out during next week
> -- I'm planning to get it attacked today during BarCamp.

Thanks for chasing this down.  Given the location, I don't think it is
even related to Kerberos, just triggered by it.

The MIT support is an important transition, and will be a key part of
the Samba 4.0 release.  I continue to be willing and able to help you
land it into master, be it for beta1 or slightly later.

Thanks,

Andrew Bartlett

-- 
Andrew Bartlett                                http://samba.org/~abartlet/
Authentication Developer, Samba Team           http://samba.org



More information about the samba-technical mailing list