# Large / slow tests, forbidding parallel execution?

**URL:** https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332
**Category:** Engineering
**Created:** [August 13, 2021, 5:11pm UTC](https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332 "2021-08-13T17:11:48Z")
**Posts on this page:** 6
**Page:** 1

<div class="post-metadata">

### Author: ![seanm](https://discourse.itk.org/letter_avatar_proxy/v4/letter/s/a88e4f/32.png) [@seanm](https://discourse.itk.org/u/seanm)
#### Post date: [August 13, 2021, 5:11pm UTC](https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332/1 "2021-08-13T17:11:48Z")

</div>

Hi all,

I recall there is a mechanism to mark ITK tests that shouldn’t run in parallel, but I don’t recall and can’t find what it is.

I’m looking to solve some of the timeouts here:

[https://open.cdash.org/viewTest.php?onlyfailed&buildid=7399376](https://open.cdash.org/viewTest.php?onlyfailed&buildid=7399376)

Some of these tests pass when run individually, but timeout when many test are run at once.

Thanks,

Sean

---

<div class="post-metadata">

### Author: ![dzenanz](https://discourse.itk.org/user_avatar/discourse.itk.org/dzenanz/32/1093_2.png) [@dzenanz](https://discourse.itk.org/u/dzenanz)
#### Post date: [August 13, 2021, 7:32pm UTC](https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332/2 "2021-08-13T19:32:19Z")

</div>

Maybe [CTEST\_PARALLEL\_LEVEL](https://cmake.org/cmake/help/latest/envvar/CTEST_PARALLEL_LEVEL.html)?

---

<div class="post-metadata">

### Author: ![dzenanz](https://discourse.itk.org/user_avatar/discourse.itk.org/dzenanz/32/1093_2.png) [@dzenanz](https://discourse.itk.org/u/dzenanz)
#### Post date: [August 13, 2021, 7:36pm UTC](https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332/3 "2021-08-13T19:36:33Z")

</div>

Or maybe [RUN\_SERIAL](https://cmake.org/cmake/help/latest/prop_test/RUN_SERIAL.html)?

---

<div class="post-metadata">

### Author: ![blowekamp](https://discourse.itk.org/user_avatar/discourse.itk.org/blowekamp/32/79_2.png) [@blowekamp](https://discourse.itk.org/u/blowekamp)
#### Post date: [August 16, 2021, 1:41pm UTC](https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332/4 "2021-08-16T13:41:14Z")

</div>

The timeout on the local system can be changed with [CTEST\_TEST\_TIMEOUT](https://discourse.itk.org/t/itk-5-1-release-candidate-1-available-for-testing/2568). Since this is a debug build it would make since to increase it from the defaults.

Also is the system running out of memory or anything during these tests? I think the TIFF test may use a lot of memory.

---

<div class="post-metadata">

### Author: ![seanm](https://discourse.itk.org/letter_avatar_proxy/v4/letter/s/a88e4f/32.png) [@seanm](https://discourse.itk.org/u/seanm)
#### Post date: [September 28, 2021, 6:29pm UTC](https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332/5 "2021-09-28T18:29:07Z")

</div>

Trying to clean up my dashboards, there are these two tests:

ITKv3DeformableRegistration8Test5  
ITKv3DeformableRegistration8Test9

which are already marked with both `RUNS_LONG` and `RUN_SERIAL` yet they don’t complete even after 16 minutes. And that’s not even with stress tools like ASan, TSan, valgrind. If I run by hand with a huge --timeout passed to ctest, they eventually pass after about 30 minutes. The machine is a dual quad-core Intel Xeon and 2.26 GHz and 28 GiB RAM. A bit old, but not horrible.

Perhaps such tests are just unreasonably long?

Sean

---

<div class="post-metadata">

### Author: ![blowekamp](https://discourse.itk.org/user_avatar/discourse.itk.org/blowekamp/32/79_2.png) [@blowekamp](https://discourse.itk.org/u/blowekamp)
#### Post date: [September 28, 2021, 6:36pm UTC](https://discourse.itk.org/t/large-slow-tests-forbidding-parallel-execution/4332/6 "2021-09-28T18:36:20Z")

</div>

These test are conditional on the “ITK\_USE\_BRAINWEB\_DATA” flag. For the valgrind and coverage build I am running this is not enabled. I believe this test data set was support to exercise more “real world” data set sizes.

Perhaps just turn off this CMake option?
