Jonn Taylor | 24 Apr 00:20 2014

Samba 4.1.7 clustering not using private dir

Using sernet 4.1 packages. I am unable to get smbd to use private dir. 
Set option in smb.conf and on command line.


         workgroup = TAYLORTELEPHONE
         realm = TAYLORTELEPHONE.COM
         netbios name = SHR01
         server string = Cluster Share
         interfaces = eth0, lo
         security = ADS
         private dir = /clusterdata/private
         log file = /var/log/samba/log.samba
         server min protocol = NT1
         client signing = if_required
         server signing = if_required
         clustering = Yes
         printcap name = /etc/printcap
         wins server =,
         template shell = /bin/bash
         winbind enum users = Yes
         winbind enum groups = Yes
         winbind use default domain = Yes
         winbind refresh tickets = Yes
(Continue reading)

Volker Lendecke | 23 Apr 08:31 2014

Re: Need for cleanup of idle winbindd child processes...

On Tue, Apr 22, 2014 at 07:20:51PM +0000, Ravindra Channabasappa wrote:
> Yes, I tested using "use mmap = no" option... still see
> the RSS size increase in the child winibndd.

I'm afraid but then I think the only realistic way to
approach this is valgrind. All other alternatives involve
very long staring at code.

The good thing is -- this problem is likely be generic enough
that if you can reproduce your winbind version on a full
Linux distro including valgrind it should show up in a
similar way and then apply to your container.

> OS info: Linux localhost2 #1 SMP Sun Apr 13
> 14:55:34 PDT 2014 i686 i686 i386 GNU/Linux
> I checked smap data. It indicates heap size increase.
> Approximately 400Kb for every 5000 user logins per run.
> I will raise a samba bug for this.

Well, while raising a bug is a good thing for this. However
-- it's the same people who listen here who will work on
this and also ask for the valgrind output :-)

> What about the patch attached? I think it can be used for
> cleanup of idle winbindd processes.

In your latest Mail I did not see a patch. Do you have a
reference to somewhere else?

(Continue reading)

Michael Adam | 22 Apr 22:46 2014

[PATCH] fix a comment typo


review / push appreciated.

Cheers - Michael
Kamen Mazdrashki | 22 Apr 02:08 2014

[PATCH] small KCC fixes

Hi Jelmer, Andrew,

Could you please review a patch of small fixes for Python KCC

Since it is more that just one commit, I've pushed it to gerrit:,n,z
Starting from:

Also attached to this mail for console-lovers :)

From 287bae01eb563d3028108c05f48823b5740b1270 Mon Sep 17 00:00:00 2001
From: Kamen Mazdrashki <kamenim <at>>
Date: Mon, 21 Apr 2014 14:32:48 +0200
Subject: [PATCH 1/6] s4:KCC: Use dsdb.DS_DOMAIN_FUNCTION_2008 constant for
 DS-Behavior comparisons

DS_BEHAVIOR_WIN2008 was used so far which is a leftover from previous
KCC implementation in "C"

Change-Id: Id9b6551073c0b17cc27e086faa315b01305f39a5
Signed-off-by: Kamen Mazdrashki <kamenim <at>>
 python/samba/       | 2 +-
 source4/scripting/bin/samba_kcc | 8 ++++----
(Continue reading)

Jeremy Allison | 21 Apr 23:02 2014

[PATCH] We should return WERR_BADFILE in _srvsvc_NetShareAdd if the path does not exist

On Mon, Apr 21, 2014 at 09:30:20AM -0700, Richard Sharpe wrote:
> Hi Jeremy,
> This patch allows Windows and things like Hyena to work more correctly:

Here is a version for master (and CC:ed to samba-technical). Let me know
if you're OK with it. Cheers,

Gustavo Zacarias | 21 Apr 15:18 2014

[PATCHv5 1/4] build: unify and fix endian tests

Unify the endian tests out of lib/ccan/wscript into wafsamba since
they're almost cross-compile friendly.
While at it fix them to be so by moving the preprocessor directives out
of main scope since that will fail.
defines separate because of different codebases.

Signed-off-by: Gustavo Zacarias <gustavo <at>>
 buildtools/wafsamba/wscript | 65 ++++++++++++++++++++++++++++++++++++++++++---
 lib/ccan/wscript            | 55 --------------------------------------
 2 files changed, 62 insertions(+), 58 deletions(-)

diff --git a/buildtools/wafsamba/wscript b/buildtools/wafsamba/wscript
index 7984227..1a2cfe6 100755
--- a/buildtools/wafsamba/wscript
+++ b/buildtools/wafsamba/wscript
 <at>  <at>  -390,9 +390,68  <at>  <at>  def configure(conf):
         conf.define('SHLIBEXT', "so", quote=True)

-    conf.CHECK_CODE('long one = 1; return ((char *)(&one))[0]',
-                    execute=True,
-                    define='WORDS_BIGENDIAN')
+    # First try a header check for cross-compile friendlyness
+    conf.CHECK_CODE(code = """#ifdef __BYTE_ORDER
+                        #define B __BYTE_ORDER
+                        #elif defined(BYTE_ORDER)
+                        #define B BYTE_ORDER
+                        #endif
(Continue reading)

Kamen Mazdrashki | 21 Apr 04:18 2014

[PATCH] samba-tool/upgrade: Fix exception thrown during upgrade from samba3

Hi Andrew, Jelmer,

this is a minor fox for exception name thrown during upgrade from samba3.

Günter Kukkukk | 18 Apr 17:06 2014

configure.developer does not find some xfs related include files

Hi all,

probably one of the WAF gurus might want to have a look.

When running ./configure.developer on opensuse-13.1 i noticed some missing
XFS related include files.
The xfs-devel package is installed.

1.) "Use xfs.h rather than libxfs.h"

2.) xfs/xqm.h is there, but the included file uuid.h cannot be found
Checking for header xfs/xqm.h
#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <stdlib.h>
#include <stddef.h>
#include <memory.h>
#include <string.h>
#include <strings.h>
#include <inttypes.h>
#include <stdint.h>
#include <unistd.h>
#include <ctype.h>
#include <stdbool.h>
#include <stdarg.h>
#include <xfs/xqm.h>
(Continue reading)

Christof Schmitt | 17 Apr 22:52 2014

[PATCH] vfs_gpfs: Avoid warnings in developer build

From 99d17843fc2481d2baa39368a8b1704e8ec4323c Mon Sep 17 00:00:00 2001
From: Christof Schmitt <cs <at>>
Date: Thu, 17 Apr 2014 13:43:53 -0700
Subject: [PATCH] vfs_gpfs: Avoid warnings in developer build

Remove an unused variable and use discard_const_p.

Signed-off-by: Christof Schmitt <cs <at>>
 source3/modules/vfs_gpfs.c |    9 ++++-----
 1 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/source3/modules/vfs_gpfs.c b/source3/modules/vfs_gpfs.c
index edbc885..a56af8f 100644
--- a/source3/modules/vfs_gpfs.c
+++ b/source3/modules/vfs_gpfs.c
 <at>  <at>  -327,7 +327,7  <at>  <at>  again:
 	*len = size;

 	errno = 0;
-	ret = smbd_gpfs_getacl((char *)fname, flags, aclbuf);
+	ret = smbd_gpfs_getacl(discard_const_p(char, fname), flags, aclbuf);
 	if ((ret != 0) && (errno == ENOSPC)) {
 		 * get the size needed to accommodate the complete buffer
 <at>  <at>  -1127,7 +1127,8  <at>  <at>  static int gpfsacl_sys_acl_set_file(vfs_handle_struct *handle,
 		return -1;

-	result = smbd_gpfs_putacl((char *)name, GPFS_PUTACL_STRUCT | GPFS_ACL_SAMBA, gpfs_acl);
(Continue reading)

Christof Schmitt | 17 Apr 22:51 2014

[PATCH] smbd: Remove unused code for dos attributes in stat struct

From fba9c5c47c650e20848bedab350bd7184e734f69 Mon Sep 17 00:00:00 2001
From: Christof Schmitt <cs <at>>
Date: Thu, 17 Apr 2014 13:46:38 -0700
Subject: [PATCH] smbd: Remove unused code for dos attributes in stat struct

This code is unused since the move to the waf build system.

Signed-off-by: Christof Schmitt <cs <at>>
 source3/smbd/dosmode.c |  139 ++----------------------------------------------
 1 files changed, 4 insertions(+), 135 deletions(-)

diff --git a/source3/smbd/dosmode.c b/source3/smbd/dosmode.c
index 831f6d0..9d0efce 100644
--- a/source3/smbd/dosmode.c
+++ b/source3/smbd/dosmode.c
 <at>  <at>  -541,115 +541,6  <at>  <at>  uint32 dos_mode_msdfs(connection_struct *conn,

- Convert dos attributes (FILE_ATTRIBUTE_*) to dos stat flags (UF_*)
-int dos_attributes_to_stat_dos_flags(uint32_t dosmode)
-	uint32_t dos_stat_flags = 0;
(Continue reading)

Jeremy Allison | 17 Apr 18:10 2014

Re: [PATCH] Fix bug 9412 - SMB2 server doesn't support recvfile.

On Thu, Apr 17, 2014 at 02:29:46PM +0800, Jones wrote:
> 2014-04-17 3:28 GMT+08:00 Jeremy Allison <jra <at>>:
>      Once the input socket is set to blocking
>      mode, then the only way splice() can
>      return EAGAIN is if the flag SPLICE_F_NONBLOCK
>      is passed into splice - which we don't
>      do.
>      So I can't currently see any way that
>      splice can return EAGAIN once the
>      input socket is set to blocking mode.
>      Are you using a custom splice() call ?
>      If not, what kernel are you using so
>      I can look at exactly the source code
>      you have ?
> Hello Jeremy,
> Yes! Im using a custom splice call,
> not just the normal kernel splice.
> Look at the sys_recvfile, 
> the flag SPLICE_F_NONBLOCK is not passed into splice,
> but the EAGAIN still returned on my testing.
> Hmm maybe this is because the custom kernel part did it,
> i would try to dig further,
> thanks you.
(Continue reading)