[Debichem-devel] Bug#1057556: Bug#1057556: elpa: FTBFS: not enough slots available

[Michael Banck]

Hi,

On Mon, Dec 11, 2023 at 02:50:40PM +0100, Santiago Vila wrote:
> El 11/12/23 a las 14:08, Michael Banck escribió:
> > So you are building elpa on a machine/vm with one cpu/vcore apparently?
> 
> No. In this case the build was made on an AWS machine of type m6a.large,
> which has 2 vcores. I'm usually including the machine type in the hostname,
> so that it may be found in the very first line of the full build log:
> 
> sbuild (Debian sbuild) 0.85.0 (04 January 2023) on m6a-large-1701732603
> 
> The specs for m6a.large machines are available here:
> 
> https://aws.amazon.com/ec2/instance-types/m6a/

Hrm, ok.
 
> > I guess it is useful information that elpa requires at least two cores
> > to executes its MPI-based testsuite
> 
> I think the end user should never have to deal with such "useful
> information".

Note that the end user is not seeing this problem, they are installing
the binary packages and are not, in general, running the test suite.
 
> If it is possible to detect beforehand that the test suite will fail
> on a given system, then the package should probably enclose
> dh_auto_test inside a suitable Makefile if/endif block.
> 
> Alternatively, if the number "2" at the end of the debian/rules
> line saying "export TEST_FLAGS=200 12 2" is the number of vcores
> requested for the tests, then the package should probably honor
> DEB_BUILD_OPTIONS=parallel=n so that no more vcores than available
> are requested.

Good point, I will investigate this.


Thanks,

Michael