Integration fail when running in parallel
GernotMaier opened this issue · 2 comments
GernotMaier commented
There is an issue with the integration tests:
- work fine when running one after another
- failed when running in parallel
e.g., try pytest -n auto --no-cov tests/integration_tests/test_ray_tracing.py
Error messages:
FAILED tests/integration_tests/test_ray_tracing.py::test_plot_image - simtools.runners.simtel_runner.SimtelExecutionError:
This is always a SimtelExecutionError
, don't understand what changed.
tobiaskleiner commented
Seems to be running fine for me in parallel:
(env) pytest -n auto --no-cov tests/integration_tests/test_ray_tracing.py
======================================== test session starts ========================================
platform linux -- Python 3.11.7, pytest-8.2.2, pluggy-1.5.0 -- /workdir/env/bin/python3.11
cachedir: .pytest_cache
Test order randomisation NOT enabled. Enable with --random-order or --random-order-bucket=<bucket_type>
rootdir: /workdir/external/simtools
configfile: pyproject.toml
plugins: cov-5.0.0, mock-3.14.0, profiling-1.7.0, random-order-1.1.1, repeat-0.9.3, xdist-3.6.1
4 workers [6 items]
scheduling tests via LoadScheduling
tests/integration_tests/test_ray_tracing.py::test_ssts[SSTS-design]
tests/integration_tests/test_ray_tracing.py::test_single_mirror
tests/integration_tests/test_ray_tracing.py::test_plot_image
tests/integration_tests/test_ray_tracing.py::test_rx
[gw2] [ 16%] PASSED tests/integration_tests/test_ray_tracing.py::test_ssts[SSTS-design]
tests/integration_tests/test_ray_tracing.py::test_integral_curve
[gw3] [ 33%] PASSED tests/integration_tests/test_ray_tracing.py::test_single_mirror
[gw1] [ 50%] PASSED tests/integration_tests/test_ray_tracing.py::test_rx
tests/integration_tests/test_ray_tracing.py::test_process_rx
[gw1] [ 66%] PASSED tests/integration_tests/test_ray_tracing.py::test_process_rx
[gw0] [ 83%] PASSED tests/integration_tests/test_ray_tracing.py::test_plot_image
[gw2] [100%] PASSED tests/integration_tests/test_ray_tracing.py::test_integral_curve
========================================= 6 passed in 4.62s =========================================
Repeated it 10 times using --random-order as well.
GernotMaier commented
Solved - it is a local configuration issue (although I don't remember that I have changed anything):
- noticed at the same as above issues that unit tests are much much slower
- need to start the container with option
--platform linux/arm64
.
Don't really understand why podman/docker suddenly decided to use linux/amd64 images (they mostly work - but not always (see above) and are way slower).