"make test" of samba master fails in a privileged Linux container

[Thomas]

On Sun, 2021-02-28 at 10:56 PM, Andrew Bartlett via samba-technical wrote:
> Patches to make 'make test' require a TESTS= parameter might be a good idea, to set expectations correctly, users wanting to at least run over the whole testsuite anyway could run with "TESTS=.".
Thank you Andrew for making this clear. I didn't find this in the documentation so if I haven't overlooked something it would probably be good to highlight the fact in the wiki or make it more prominent.

Now I need a little more help please.
Let me first explain what my objective of running "make test" is.

I did patch the samba-tool provisioning script to enable the provisioning of an AD DC with the nfs4acl_xattr VFS module instead of the acl_xattr module.
I can now provision the AC DC without errors and also run sysvolcheck and sysvolreset without errors.
My plan was to use "make test" as a next step to check this setup and see what bombs get dropped on me. :-)

After superficially skimming through the code for "make test" it looks like I will have to modify some of the test environment smb.conf files at their " vfs objects" definition to make them use the nfs4acl_xattr VFS module.
In parallel I would like to run the tests with the standard setup for comparison. 

What would be the relevant initial targets to use in the TEST= parameter for testing the basic AD DC functionality ?

  -------
Thanks
Thomas

> I did clone the samba git repo within a privileged Debian 10 LXC.
> "configure" and "make" completed successful but "make test" fails
> with:
> 
> FAILED (183 failures, 0 errors and 2 unexpected successes in 63
> testsuites)
> 
> It is an unmodified build of samba master. What am I missing ?
> Is it not possible to successfully run "make test" in a privileged 
> Linux container ?

I think you will find that 'make test' fails also on a full VM, but if you like you can check if you want to spend the time and CPU cycles.

Untested code is broken code and these days 'make test' is only run in chunks as part of our autobuild and GitLab CI process, so it is no shock to learn that it fails as a whole.  

I'm pretty sure it has been like that for a while.

Patches to make 'make test' require a TESTS= parameter might be a good idea, to set expectations correctly, users wanting to at least run over the whole testsuite anyway could run with "TESTS=.".

Sorry,

Andrew Bartlett

-- 
Andrew Bartlett (he/him)       https://samba.org/~abartlet/
Samba Team Member (since 2001) https://samba.org
Samba Team Lead, Catalyst IT   https://catalyst.net.nz/services/samba

Samba Development and Support, Catalyst IT - Expert Open Source Solutions