[openstack-dev] [manila][puppet-manila] Bypassing jenkins FAILURE

sac sacchi at gmail.com
Fri Jul 3 02:37:32 UTC 2015


Hi,

On Thu, Jul 2, 2015 at 11:12 PM, Jeremy Stanley <fungi at yuggoth.org> wrote:

> On 2015-07-02 22:00:43 +0530 (+0530), sac wrote:
> > I have a patch[1] which should work only on Red Hat systems and
> > fails on non Red Hat systems due to non-availability of a package.
>
> I suppose it's up to the puppet-manager-core group members to decide
> whether non-portable additions are something they want to support.
>
>
Agree. This currently supports only Red Hat because there is no
corresponding package for nfs-ganesha in Debian/Ubuntu.


> > However, jenkins runs the beaker tests on Ubuntu and by definition
> > this fails, which is quite intended. However, jenkins marks
> > FAILURE due to this error.
> >
> > Due to this, patch stays in Verified-1 state incorrectly.
>
> The job has been instructed to run against the entirety of the
> acceptance spec, and the contents of that spec aren't currently
> conditional for specific classes (besides repo selection and package
> providers).
>
>
Okay, makes sense.


> > How to bypass or waive this failure? Can I request an exception?
> >
> > [1] https://review.openstack.org/#/c/186699/
>
> Waiving your change past the job wouldn't solve this, since that job
> would just become instantly broken for the puppet-manila repo from
> that point on (and also we have no way to actually do that short of
> disabling the job or involving an admin for the code review system
> itself to bypass our gating safeties). If
> non-portable/platform-specific features are acceptable to the
> puppet-manila maintainers, I think your options are to either wrap
> your addition to the spec in a check for $::osfamily == 'RedHat' or
> modify your new class to not fail when applied on non-RedHat-family
> platforms.
>

I think this is a good plan. Instead of `fail', I would go ahead with
`warning'. Thanks.


>
> Anyway, I suspect this is a discussion far better had in the change
> review itself rather than on the mailing list.
>

Sure.

Thanks,
sac.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150703/dbbc83da/attachment.html>


More information about the OpenStack-dev mailing list