Re: Updating CI-runners to buster

[Henning Schild <henning.schild@siemens.com>]

On Wed, 11 Mar 2020 11:37:30 +0100
Baurzhan Ismagulov <ibr@radix50.net> wrote:

> On Wed, Mar 11, 2020 at 09:28:29AM +0100, Henning Schild wrote:
> > > > i think we can now enable the two arm64 CI jobs that are
> > > > currently commented out to cater for old CI runners.    
> > > 
> > > Makes sense. Then, I'd suggest to drop stretch host support.
> > > Opinions?  
> > 
> > What does dropping mean? I would keep it alive as long as possible,
> > especially while stretch is still alive. For the CI runners and all
> > possible targets i.e. buster arm64 we can drop it.  
> 
> By "dropping stretch host support" I meant (option 1):
> 
> 1. Remove Jenkins jobs building on stretch:
> 
>    1. isar_master_stretch
>    2. isar_master_stretch_fast
>    3. isar_next_stretch
>    4. isar_next_stretch_fast
> 
> 2. Keep building all targets as defined in scripts/ci_build.sh next
> ad81c98. Add qemuarm64-buster and qemuarm64-bullseye.
> 
> 
> I understand your suggestion as follows (option 2):
> 
> 1. Keep Jenkins stretch builders (as above,
> {master,next}_stretch[_fast]).
> 
> 2. Add qemuarm64-{buster,bullseye} targets for buster builders but
> not for the stretch ones.
> 
> Is this what you mean?
> 
> 
> I would support proceeding with option #2.

Seems you are just thinking about which combinations to actually test
in CI. I think dropping stretch hosts from the official testing is no
problem.
You said "dropping stretch host support" which to me sounded like
actually removing code or allowing code that means hosts will have to
be >=buster.

Dropping tests is OK, dropping support is not. Of cause testing ensures
support, but i would hope that the risk of a breakage is low.

Henning

> 
> With kind regards,
> Baurzhan.
>