(not) breaking GitLab CI

Andrew Bartlett abartlet at samba.org
Wed Oct 31 20:44:27 UTC 2018


On Wed, 2018-10-31 at 08:50 -0700, Jeremy Allison via samba-technical
wrote:
> On Wed, Oct 31, 2018 at 10:10:46AM +0100, Andreas Schneider via
> samba-technical wrote:
> > 
> > I've tried to update nss_wrapper in Samba but it fails.
> > 
> > https://gitlab.com/samba-team/devel/samba/pipelines/34920284
> > 
> > The strange thing is that it works perfectly fine if I run it
> > locally. I will 
> > try to reproduce it in the docker image.
> Does it fail on sn-devel ? sn-devel is "truth", the gitlab error
> reports
> are still more of a matrix-like reality :-).

I agree, it is a matrix-like reality, but as seen today it is a pretty
good matrix. 

Even so, given the broad team adoption it would serve nobody to merge
patches that break the GitLab CI system which we all rely on.  

Doing so would just move the pressure to a different subset of the
team, not resolve the issue.  Therefore I hope we can all agree such
actions, stopping the work of everyone who uses GitLab for CI system
daily, would be decidedly anti-social. 

Think of it like a flapping test.  Yes, you can get it past the
machine, but no: at this point breaking GitLab is not OK.

Of course we should continue to make our build environments as similar
as possible, because we now have many developers without access to sn-
devel that need reliable build results.  (Perhaps some day sn-devel can
run the same docker image, or even trigger a GitLab CI run?)

Thanks,

Andrew Bartlett

-- 
Andrew Bartlett
https://samba.org/~abartlet/
Authentication Developer, Samba Team         https://samba.org
Samba Development and Support, Catalyst IT   
https://catalyst.net.nz/services/samba







More information about the samba-technical mailing list