[Samba] self compiled 4.10.3 replication failure.

Rowland penny rpenny at samba.org
Tue May 21 16:44:53 UTC 2019


On 21/05/2019 17:06, me at tdiehl.org wrote:
> On Tue, 21 May 2019, Rowland penny via samba wrote:
>
>> On 21/05/2019 01:11, Tom Diehl via samba wrote:
>>>
>>>  Here's where I disagree. When you run ./configure, make and make 
>>> install
>>>  everything gets put into /usr/local/samba by default. It is not 
>>> installed
>>>  over top of any system components.
>>>
>>>  Not that I recommend this as a standard practice but if you totally 
>>> screw
>>>  the
>>>  pooch on a build for whatever reason all you have to do is
>>>  rm -r /usr/local/samba and you get to start over. I have tested 
>>> this and
>>>  know
>>>  that it works. :-)
>>>
>>>  Obviously if you do that to a DC that is joined to a domain, you 
>>> have more
>>>  than
>>>  that to clean up.
>>>
>>>  Regards,
>>>
>> Whilst you can just run './configure' and have everything put into 
>> /usr/local/samba, if RHEL is like Debian, this comes with a couple of 
>> problems. You have to ensure that '$PATH' points to /usr/local/samba 
>> first and various things still expect to find Samba from the distro 
>> packages (gvfs is one, if I remember correctly).
>
> The 2 things that I have found that I need to set outside of 
> /usr/local/samba
> are the $PATH and 2 symlinks for winbind. Both of these are documented 
> on the
> wiki (Thank you). If you use some type of configuration management 
> such as ansible to
> build the DCs you do not even have to think about setting these. Other 
> than
> that I have not needed to do anything else for a DC. For a file server 
> I use
> the distro supplied packages.
>
> I do not know about the gvfs stuff. Since I only build dedicated DCs.
>
>> The problem, as I see it, is that RHEL is a bit late to the party and 
>> is where Debian was 5 years ago. On top of this is the extra problem 
>> of python3.
>
> RHEL has always been behind as far as an AD DC is concerned because of 
> the
> MIT BS.
>
>> Once methods to build Samba packages on RHEL are learnt, we will 
>> probably look back and ask 'what was the problem' ;-)
>
> Having read this list for the last couple of years, I am starting to 
> wonder
> why people are making this all so hard. I understand that pre-built 
> packages are
> most desirable. I am a firm believer of that but in the case of a 
> samba AD DC building rpms seems like a lot of extra work for little 
> return.
>
> If I did not have something like ansible to allow me to make the 
> builds repeatable
> and automated I might think differently but this works for me. :-)
>
> Rowland, any chance of getting
> https://wiki.samba.org/index.php/Package_Dependencies_Required_to_Build_Samba#Red_Hat_Enterprise_Linux_7_.2F_CentOS_7_.2F_Scientific_Linux_7 
>
> updated to reflect the package differences required for 4.10?

There is every chance, once I find out just what the required packages 
are ;-)

There was some talk of this happening automatically, but this seems to 
have not come to fruition yet.

>
> I simply enabled the epel-repository and substituted python3x-devel for
> python-devel and added python3x-dns.
>
> The python3x-dns package was not obvious but without it the build 
> would succeed
> but the samba-tool dns module would not work.
>
> I kinda thought the configure script would complain if I was missing 
> some of
> the required python3 bits but it did not.
>
You and me both, I think the problem is that using python3 is so new 
that problems like this are bound to crop up because nobody realised 
they needed fixing.
> Is this a bug in the configure script or is this expected?
>
Not sure, it sort of depends on what it checks for, if it is supposed to 
check for python3.x which is required for the build and it isn't found 
and the compile succeeds, then it probably is a bug. If however the 
missing component is something required by a script that is only run on 
a running system, then it isn't really a bug, it is a lack of a dependency.

Rowland




More information about the samba mailing list