WIP: Continuous integration of Slicer on centos (updated & rebased)

[MayeulChassagnard]

Update of https://github.com/Slicer/Slicer/pull/607

[jcfr]

After manually fetching your changes into /tmp/test/Slicer

I tweaked my small script to work for existing checkout. It is currently building.

TEST_DIR=/tmp/test
mkdir -p $TEST_DIR

cd $TEST_DIR
dir=docker-opengl-centos
[[ ! -e $dir ]] && git clone git://github.com/thewtex/docker-opengl -b centos $dir
cd $dir
make

cd $TEST_DIR
dir=SlicerDocker
[[ ! -e $dir ]] && git clone git://github.com/MayeulChassagnard/SlicerDocker -b slicer-build-with-tests-rebased $dir
cd $dir
./build.sh

cd $TEST_DIR
dir=Slicer
[[ ! -e $dir ]] && git clone git://github.com/MayeulChassagnard/Slicer -b Continuous-Integration-on-centos-2016.10.20 $dir
cd $dir
./CMake/CircleCI/build.sh
./CMake/CircleCI/run.sh

[MayeulChassagnard]

@jcfr @thewtex What is the status now? Can we merge this to be able to know finally if the all process works or not? During that time, I am going to try to solve the few remaining test error

[thewtex]

@MayeulChassagnard The process status can be determined by the results on this PR.

There are many people in the community collaboratively developing Slicer. It is important not to merge code with known build errors or test failures so that others are not distracted or confused by them.

[MayeulChassagnard]

Wooooww! :fireworks: :raised_hands: :tada: It works finally! CircleCI details

Going to update SlicerDocker to run all tests!

[MayeulChassagnard]

We are facing a major problem..

Because test's results are displayed only at the end and it takes more than 10 minutes to run all these Slicer's tests, CircleCI is running into a Timed Out.

[jcfr]

What about running the test with -V or even -VV ?

[MayeulChassagnard]

It is already done: https://github.com/thewtex/SlicerDocker/blob/master/slicer-test/opengl/test.sh#L19

The fact is that it'll be displayed only when the container is stopped: https://github.com/MayeulChassagnard/Slicer/blob/Continuous-Integration-on-centos-2016.10.27-rebased/CMake/CircleCI/run_opengl.sh#L135-L148

[ihnorton]

@MayeulChassagnard some possible options:

• override the timeout
• send some (subset) of output to syslog and use docker logs --follow ... to get it (replacing docker wait).
• or simplest might be to do:

while docker inspect -f {{.State.Running}} $container
do
  sleep(5)
  echo -n .
done

[MayeulChassagnard]

py_AtlasTest seems not stable (http://slicer.cdash.org/index.php?project=Slicer4&date=2017-01-09##Experimental)

Should we remove it from the test list?

[pieper]

Is it just a timeout issue?

On Thu, Jan 12, 2017 at 12:41 PM, Mayeul Chassagnard < [email protected]> wrote:

py_AtlasTest seems not stable (http://slicer.cdash.org/ index.php?project=Slicer4&date=2017-01-09##Experimental)

Should we remove it from the test list https://github.com/thewtex/SlicerDocker/blob/master/slicer-test/opengl/CircleCI_Slicer_Docker.cmake#L68 ?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/Slicer/Slicer/pull/609#issuecomment-272230493, or mute the thread https://github.com/notifications/unsubscribe-auth/AAHsfWUMCJw91bxlmNvwCFl0gsB69-7Zks5rRmXMgaJpZM4KioiR .

[MayeulChassagnard]

@pieper Yes, and note that when I manually rebuild (by clicking the 'rebuild' button in CircleCI), the test becomes green (because it's executed in less than 10 minutes)