[petsc-dev] IMPORTANT nightly builds are down do not add to next

Jed Brown jed at jedbrown.org
Sat Dec 30 17:34:27 CST 2017


Jeff Hammond <jeff.science at gmail.com> writes:

> On Sat, Dec 30, 2017 at 12:04 PM Jed Brown <jed at jedbrown.org> wrote:
>
>> "Smith, Barry F." <bsmith at mcs.anl.gov> writes:
>>
>> >> On Dec 30, 2017, at 3:53 AM, Matthew Knepley <knepley at gmail.com> wrote:
>> >>
>> >> On Fri, Dec 29, 2017 at 5:35 PM, Smith, Barry F. <bsmith at mcs.anl.gov>
>> wrote:
>> >> > On Dec 29, 2017, at 3:52 PM, Matthew Knepley <knepley at gmail.com>
>> wrote:
>> >> >
>> >> > On Fri, Dec 29, 2017 at 2:11 PM, Smith, Barry F. <bsmith at mcs.anl.gov>
>> wrote:
>> >> >
>> >> >    The nightly builds are down (out-dated data at the website) and
>> Satish is in India and cannot fix it.
>> >> >
>> >> >     Please do not put anything into next until you are notified that
>> the nightly tests are working again.
>> >> >
>> >> > Cool.
>> >> >
>> >> > This should tell us that we need to make more of an effort to
>> automate this part, and employ failsafe measures.
>> >>
>> >>   Indeed. Note that this is largely just bad timing, this happens very
>> rarely and it is just bad luck Satish is out.
>> >>
>> >> I agree completely. However, right now we are all too indispensible. We
>> usually focus on transmitting developer knowledge,
>> >> but I think we need more focus on transmitting process knowledge. Could
>> another group of people pick up and run
>> >> the maintenance process for PETSc?
>> >
>> >    No because they would use jenkins or some similar worthless shit;
>> >    we need to keep control of the processes but need more more help in
>> >    running the processes.
>>
>> It shouldn't be this labor-intensive.  As I understand it, we could use
>> free hosted CI except for proprietary compilers/unsupported
>> environments.  Most of those CI systems want to sell their products with
>> enterprise pricing for custom environments, but I think it's supposed to
>> be easy to use the gitlab-ci runner on a custom machine.
>>
>
> You can use ICC and PGI in Travis CI, which is free if you use Github with
> a public repo. ICC requires a friend at Intel to help with license issues.
> PGI has community edition you should be able to use. Travis CI doesn’t
> support Bitbucket but I recall you mirror Petsc there. Examples of both are
> easily found with an obvious google search.

Good to know.

> Travis CI is great for covering generic Mac/Homebrew and Ubuntu platforms.
> You can get almost all compiler versions and MPI libraries in both, with
> reasonable effort.
>
> I can do an initial implementation if you aren’t inclined to do it
> yourselves.

Lisandro set up Travis-CI last year, though we aren't currently doing
pull requests there.

https://travis-ci.org/petsc/petsc/branches

Our "weird" configurations are OpenSolaris, FreeBSD, and MS Windows.  I
see some murmurs online about running QEMU inside Travis-CI, but it
isn't supported.  It's perhaps a bit hokey, but a Travis-CI build could
launch an EC2 Windows instance (or connect to a Windows machine at
Argonne) and stream the results back.


More information about the petsc-dev mailing list