Boost logo

Boost Testing :

From: Tom Kent (lists_at_[hidden])
Date: 2019-06-20 11:36:40


On Fri, Jun 14, 2019 at 12:58 PM Tom Kent <lists_at_[hidden]> wrote:

>
>
> On Fri, Jun 14, 2019 at 11:20 AM Tom Kent <lists_at_[hidden]> wrote:
>
>>
>>
>> On Fri, Jun 14, 2019 at 10:13 AM Rene Rivera via Boost-Testing <
>> boost-testing_at_[hidden]> wrote:
>>
>>>
>>>
>>> On Fri, Jun 14, 2019 at 9:50 AM Tom Kent via Boost-Testing <
>>> boost-testing_at_[hidden]> wrote:
>>>
>>>>
>>>>
>>>> On Fri, Jun 14, 2019 at 9:14 AM Rene Rivera via Boost-Testing <
>>>> boost-testing_at_[hidden]> wrote:
>>>>
>>>>> On Fri, Jun 14, 2019 at 8:25 AM Tom Kent via Boost-Testing <
>>>>> boost-testing_at_[hidden]> wrote:
>>>>>
>>>>>> For the last few days (Since June 4?) I've been having problems with
>>>>>> all my linux tests running against the develop branch.
>>>>>>
>>>>>
>>>>> Are you running the tests in a Linux container?
>>>>>
>>>>>
>>>> Yes.
>>>>
>>>> Container images created from here:
>>>> https://github.com/teeks99/boost-cpp-docker
>>>>
>>>
>>> Then the likely issue is a change to b2 that defaults to using parallel
>>> jobs to match the available system cores and launches so many compile
>>> commands that it runs out of memory. It seems most container virtualization
>>> systems are terrible at virtualizing the system information and leak the
>>> host information. Only immediate option is to pass b2 an appropriate "-jN"
>>> option. Or wait until I fix the problem (I already tried one solution, will
>>> keep trying).
>>>
>>>
>> Hmm, I'm not sure that is it. I was already running with -j8, on a 8 core
>> machine.
>>
>> I run:
>>
>> run.py --runner=teeks99-dkr-dc8-2a --toolsets=clang-8~c++2a \
>> "--bjam-options=-j8 address-model=64 --remove-test-targets" \
>> --comment=info.html --tag=develop
>>
>>
>>
>
> I also have one physical machine (arm based) that is running the
> same(-ish) thing, with the same failure.
>

Any luck tracking this down? I haven't been able to figure out a mitigation
from my side.

If the changes on develop are causing problems like this, can we revert
them until we figure it out?

Tom



Boost-testing list run by mbergal at meta-comm.com