branch inclusion request

branch inclusion request

[Sasha Levin]

Hi all,

Right now I'm pushing stable patches to Greg for his weekly releases,
and for those branches I only do basic cross-compilation and no
boot/functional testing at all.

I'd like to modify my process to rely on KernelCI testing instead. This
means I'd like to include my branches (one for each stable kernel) into
the build system.

These branches get pushed a few times per week, and sometimes a few
times a day, so this will generate a fair amount of builds. OTOH, I
don't really need to keep the build results for too long as I try to
address any issues that come up the same day.

--
Thanks,
Sasha

Re: [kernelci] branch inclusion request

[Guillaume Tucker]

[-- Attachment #1: Type: text/plain, Size: 1547 bytes --]

Hi Sasha,

On Fri, Oct 26, 2018 at 5:08 PM Sasha Levin <sashal@kernel.org> wrote:

> Hi all,
>
> Right now I'm pushing stable patches to Greg for his weekly releases,
> and for those branches I only do basic cross-compilation and no
> boot/functional testing at all.
>
> I'd like to modify my process to rely on KernelCI testing instead. This
> means I'd like to include my branches (one for each stable kernel) into
> the build system.
>
> These branches get pushed a few times per week, and sometimes a few
> times a day, so this will generate a fair amount of builds. OTOH, I
> don't really need to keep the build results for too long as I try to
> address any issues that come up the same day.
>

This sounds like a good idea in principle, but what is the actual
gain given the extra load it would take to build and test all
these branches?

Of course it's very important to test stable releases, but they
typically show very few build or test failures and there is
already stable-rc to catch any of these.  It seems to me that
development branches are where extra testing efforts would be
required the most, to catch issues earlier.

If adding your stable branches to KernelCI would save you a lot
of manual effort, then yes I guess that would probably be a good
reason.  If you'll still be doing the same manual builds and
checks before pushing your branches, then I'm not really sure
it's worth adding them to KernelCI as it currently stands.

Let's see what others think the priorities should be wrt adding
more branches.

Guillaume


>
>

[-- Attachment #2: Type: text/html, Size: 2330 bytes --]

Re: [kernelci] branch inclusion request

[Kevin Hilman]

On Fri, Oct 26, 2018 at 9:35 AM Guillaume Tucker
<guillaume.tucker@gmail.com> wrote:
>
> Hi Sasha,
>
> On Fri, Oct 26, 2018 at 5:08 PM Sasha Levin <sashal@kernel.org> wrote:
>>
>> Hi all,
>>
>> Right now I'm pushing stable patches to Greg for his weekly releases,
>> and for those branches I only do basic cross-compilation and no
>> boot/functional testing at all.
>>
>> I'd like to modify my process to rely on KernelCI testing instead. This
>> means I'd like to include my branches (one for each stable kernel) into
>> the build system.
>>
>> These branches get pushed a few times per week, and sometimes a few
>> times a day, so this will generate a fair amount of builds. OTOH, I
>> don't really need to keep the build results for too long as I try to
>> address any issues that come up the same day.
>
>
> This sounds like a good idea in principle, but what is the actual
> gain given the extra load it would take to build and test all
> these branches?
>
> Of course it's very important to test stable releases, but they
> typically show very few build or test failures and there is
> already stable-rc to catch any of these.  It seems to me that
> development branches are where extra testing efforts would be
> required the most, to catch issues earlier.
>
> If adding your stable branches to KernelCI would save you a lot
> of manual effort, then yes I guess that would probably be a good
> reason.  If you'll still be doing the same manual builds and
> checks before pushing your branches, then I'm not really sure
> it's worth adding them to KernelCI as it currently stands.
>
> Let's see what others think the priorities should be wrt adding
> more branches.

IMO, since we're trying to make stable rock solid, this is a good
addition.  Also, since Sasha is also helping with build horsepower, I
think we should add these trees.

Kevin

Re: [kernelci] branch inclusion request

[Guillaume Tucker]

[-- Attachment #1: Type: text/plain, Size: 2201 bytes --]

On Mon, Oct 29, 2018 at 3:35 PM Kevin Hilman <khilman@baylibre.com> wrote:

> On Fri, Oct 26, 2018 at 9:35 AM Guillaume Tucker
> <guillaume.tucker@gmail.com> wrote:
> >
> > Hi Sasha,
> >
> > On Fri, Oct 26, 2018 at 5:08 PM Sasha Levin <sashal@kernel.org> wrote:
> >>
> >> Hi all,
> >>
> >> Right now I'm pushing stable patches to Greg for his weekly releases,
> >> and for those branches I only do basic cross-compilation and no
> >> boot/functional testing at all.
> >>
> >> I'd like to modify my process to rely on KernelCI testing instead. This
> >> means I'd like to include my branches (one for each stable kernel) into
> >> the build system.
> >>
> >> These branches get pushed a few times per week, and sometimes a few
> >> times a day, so this will generate a fair amount of builds. OTOH, I
> >> don't really need to keep the build results for too long as I try to
> >> address any issues that come up the same day.
> >
> >
> > This sounds like a good idea in principle, but what is the actual
> > gain given the extra load it would take to build and test all
> > these branches?
> >
> > Of course it's very important to test stable releases, but they
> > typically show very few build or test failures and there is
> > already stable-rc to catch any of these.  It seems to me that
> > development branches are where extra testing efforts would be
> > required the most, to catch issues earlier.
> >
> > If adding your stable branches to KernelCI would save you a lot
> > of manual effort, then yes I guess that would probably be a good
> > reason.  If you'll still be doing the same manual builds and
> > checks before pushing your branches, then I'm not really sure
> > it's worth adding them to KernelCI as it currently stands.
> >
> > Let's see what others think the priorities should be wrt adding
> > more branches.
>
> IMO, since we're trying to make stable rock solid, this is a good
> addition.  Also, since Sasha is also helping with build horsepower, I
> think we should add these trees.
>

Sounds like the overal consensus is to enable them.

Sasha, could you please provide the list of the branch names and
send a PR with the URL of your tree in kernelci-core-staging?

Guillaume

[-- Attachment #2: Type: text/html, Size: 3037 bytes --]