Stefan (metze) Metzmacher | 31 Oct 05:59 2014

Re: Current SMB2 leases patchset (WIP).

Am 30.10.2014 um 20:10 schrieb Jeremy Allison:
> On Thu, Oct 30, 2014 at 04:33:49PM +0100, Stefan (metze) Metzmacher wrote:
>> Hi Jeremy,
>> one more...
>>> here's a first independent patchset set that can go to master now.
>>> Mostly the strict rename patches.
>>> The rest of the patches is in the following branch:
>>> I'll post more review details on the rest once I looked at it more closely.
>>> metze
> OK, I've pushed the previous patchset + this one
> to master.
> Attached is the rebase of my leases-wip patches
> remaining on top, just so I can keep working on
> it and ensure the logic hasn't changed :-).
> I'll accept more fixes to push as you get done
> with them !
> Thanks for doing this metze. I appreciate the
> work you're putting into it (and I noticed and
> adopted a fix you already did in your wip tree,
> removing the superfluous del_share_mode()
(Continue reading)

Kamen Mazdrashki | 31 Oct 04:08 2014

[MS-ADTS] Undelete Processing Specifics Clarification

Dear Dochelp team,

I am currently working on Tombstone reanimation implementation in Samba
and I am having troubles to understand how objectCategory attribute should
be processed.

According to
objectCategory attribute should be restored to its default value in case
it is *not* specified by the user. I guess this mean it is not specified in
the special ldap modify request.

I am testing against Windows Server 2008 R2 with Forest Functional level
"2008 R2"

   1. when objectCategory is not specified, everything is fine
   2. when objectCategory is specified though, I am always getting
   LDAP_UNWILLING_TO_PERFORM error. I have tried both to
   "reaplace" and "add" this attribute - same result. Please see attached

My question is: how to specify this attribute so I am able to control the

Best Regards,
Kamen Mazdrashki
Attachment (reanimate-w2k8.ldif): text/x-ldif, 453 bytes
Jeremy Allison | 31 Oct 00:54 2014

Re: Current SMB2 leases patchset (WIP).

On Thu, Oct 30, 2014 at 12:10:13PM -0700, Jeremy Allison wrote:
> On Thu, Oct 30, 2014 at 04:33:49PM +0100, Stefan (metze) Metzmacher wrote:
> > Hi Jeremy,
> > 
> > one more...
> > 
> > > here's a first independent patchset set that can go to master now.
> > > Mostly the strict rename patches.
> > > 
> > > The rest of the patches is in the following branch:
> > >;a=shortlog;h=refs/heads/master3-leases
> > > 
> > > I'll post more review details on the rest once I looked at it more closely.
> > > 
> > > metze
> OK, I've pushed the previous patchset + this one
> to master.

Which fails because the Samba4 file server doesn't


I'll add to knownfail for that and repush.

Arggh. Time to disable that fileserver I think.

Richard Sharpe | 30 Oct 23:49 2014

Do we know about this deprecation warning for the md5 module in waf?

Hi folks,

I see this when building 4.1.12:

./buildtools/wafsamba/ DeprecationWarning: the md5
module is deprecated; use hashlib instead

Should I prepare a patch or has this been fixed already upstream or whatever?


Richard Sharpe

Marc Muehlfeld | 30 Oct 21:42 2014

[PATCH] samba-tool: Create NIS enabled users and unixHomeDirectory attribute


a few weeks ago I submitted a patch that allows samba-tool to create NIS
enabled groups (it's already in master) . Now I was asked on the Samba
Mailing list to provide this feature for user accounts, too.

Example usage:
# samba-tool user add demo30 password --nis-domain=samdom
--unix-home=/home/demo30 --uid-number=13030 --login-shell=/bin/bash

If --nis-domain was given, then the other four parameters are mandatory
(like in ADUC). If it's not set, then they are optional and
all/some/none of them can be set, like before. I'm not sure, if this
makes sense, but it would not break existing scripts from users if e. g.
just some of this four attributes are given.

Attached is the patch. Please review and push, if OK.

Richard Sharpe | 30 Oct 20:17 2014

Why does RedHat require that iniparser-devel be installed to build their 4.1.1 RPM when the source is in the tree?

Hi folks,

Does anyone have a clue why they have iniparser-devel as a
BuildRequire when the code is in the source tree they import?


Richard Sharpe

Andreas Schneider | 30 Oct 16:12 2014

[PATCH 0/4] Fix uninitialized variables.

These patches fix some uninitialized variables. cppcheck complained
about them.

Andreas Schneider (4):
  s3-rpcclient: Make sure current_nt_hash is initialized.
  s3-libsmb: Make sure the stat structure is initialized.
  torture: Make sure server_type isn't used uninitialized.
  libcli: Make sure status is initialized.

 libcli/cldap/cldap.c            | 2 +-
 source3/libsmb/clirap.c         | 2 +-
 source3/rpcclient/rpcclient.c   | 4 +++-
 source4/torture/ldap/netlogon.c | 2 +-
 4 files changed, 6 insertions(+), 4 deletions(-)



Richard Sharpe | 30 Oct 03:58 2014

Patch: Don't drop any error codes when user asks for a level 10 log

Hi folks,

Attached is a small patch to ensure that in the auth path we don't
drop error codes when the user asked for a level 10 log. It makes it
hard for admins to diagnose problems if we elide info like that.


Richard Sharpe
David Disseldorp | 30 Oct 01:37 2014

[PATCH 0/6] Fix spoolss EnumJobs and GetJob request handling

This patch set attempts to address bug 10905, which is nicely reproduced
by the existing rpc.spoolss.printer.addprinter.print_test_extended
smbtorture test.

Currently EnumJobs and GetJobs spoolss request handling is broken, in
that Samba uses and returns printer back-end allocated job (sysjob)
identifiers, rather than those allocated by spoolss and returned on job

I have one further patch which ensures that print_test_extended is not
ignored during selftest (as it now passes), but I'd like to hold off on
it until I can find out what is causing the flakiness in one of the
other spoolss tests.

Feedback welcome.

Cheers, David


 source3/include/printing.h                  |   1 +
 source3/printing/printing.c                 |  70 +++++++++++++++++++++++++++++++-----------
 source3/rpc_server/spoolss/srv_spoolss_nt.c | 212 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---------------------------------------------
 3 files changed, 190 insertions(+), 93 deletions(-)

Richard Sharpe | 29 Oct 23:45 2014

Which version of CTDB should be used with Samba 4.1.12?

Hi Folks,

Which one should I use?

Richard Sharpe
ray klassen | 29 Oct 23:13 2014

Samba Upgrade-iad

First of all let me congratulate the wiki writers. The step by step classic-upgrade guide is very helpful.
Here are my notes on the various steps of the upgrade. 

-- created a vanilla debian wheezy install, installed all the prerequisites as well as "devscripts," 
--compiled, installed samba using samba-4.1.2 
-- created symbolic links from /usr/local/samba/bin to /usr/local/bin and /usr/local/samba/sbin to
/usr/local/sbin because those directories are in $PATH and from /usr/local/samba/etc/ to /etc/samba
and from /usr/local/samba/var/log.* to /var/log/samba/* so that those files will be where I expect.
-- installed slapd, copied over the current ldap files, configured slapd to load them-- copied smb.conf
and various *db files to a directory-- downloaded the debian bind9 source deb, added 
--with-dlopen=yes to EXTRA_FEATURES= in the debian/rules file
--ran debuild -us -uc  from bind9 source dir -- created debs with dlopen support (this is what devscripts
was for. 
--ran samba-tool doman classicupgrade... with --dns-backend=BIND_DLZ etc.--several colisions had to
be edited out of the ldap directory before the upgrade would complete         -- a trusted
domain account had to be removed-- an early phase of the classicupgrade script warned me that it would not
be imported, but a later phase choked apparently because it hadn't been imported. Bug?        
-- two groups had different groupnames but the same DisplayName. that had to be changed.
 -- played around with dns. Found that windows boxes really like to talk to the domain controller itself
and not a slave.

-- have been busy reconnecting all the services that depended on ldap to active directory, learning kerberos

Some things did not work as expected. 1) all the computers did not automatically join the new domain. Some
did and some did not. The computers that were at the head office presumably in the same broadcast domain all
joined automatically, once I configured the domain controller as DNS server assigned by DHCP. The
computers at our satellite offices (approximately 30) did not. This maybe because I had LMHOSTS files on
all those machines, except that after delete and reboot, (DNS still pointed at the DC -- I didn't forget)
(Continue reading)