Fix flaky sandbox_options test

[MaciejKaszynski]

Description

It seems the //tests/integration/sandbox_options:sandbox_options test is pretty flaky.
https://github.com/eclipse-score/lifecycle/actions/runs/32465690328/job/96721678652

Analysis results

No response

Solution

No response

Error Occurrence Rate

None

How to reproduce

No response

Supporting Information

No response

Classification

Minor

First Affected Release

not released (main)

Last Affected Release

not released (main)

Expected Fixed Release

before release (main)

Category

• Safety Relevant
• Security Relevant

[TimoSteuerwaldETAS]

I had a look onto it. It seems that sometimes the shutdown in launch manager takes far too long.
For example:

[2026-08-21 09:02:23.619] [INFO] [launch_manager] [  PASSED  ] 1 test.
[2026-08-21 09:02:23.622] [INFO] [launch_manager] 2026/08/21 09:02:23.2943622 2353113 000 ECU1 LM LM log debug verbose 10 Process 2 pid 85 ( sandbox_options_process_c ) for node True terminated with status 0
[2026-08-21 09:02:23.622] [INFO] [launch_manager] 2026/08/21 09:02:23.2943622 2353114 000 ECU1 LM LM log debug verbose 5 Component  2  finished  activation  successfully
[2026-08-21 09:02:23.668] [INFO] [launch_manager] 2026/08/21 09:02:23.2943667 2353570 000 ECU1 LM LM log debug verbose 4 Process with Id sandbox_options_process_c received supervision event 0
[2026-08-21 09:02:24.096] [DEBUG] [tests.utils.testing_utils.run_until_file_deployed] Waiting for /tmp/tests/test_end
[2026-08-21 09:02:24.654] [DEBUG] [tests.utils.testing_utils.run_until_file_deployed] Waiting for /tmp/tests/test_end
[2026-08-21 09:02:25.198] [DEBUG] [tests.utils.testing_utils.run_until_file_deployed] Waiting for /tmp/tests/test_end
[2026-08-21 09:02:25.754] [DEBUG] [tests.utils.testing_utils.run_until_file_deployed] Waiting for /tmp/tests/test_end
[2026-08-21 09:02:26.325] [DEBUG] [tests.utils.testing_utils.run_until_file_deployed] Waiting for /tmp/tests/test_end
[2026-08-21 09:02:26.985] [INFO] [launch_manager] 2026/08/21 09:02:26.2946984 2386733 000 ECU1 LM LM log debug verbose 1 Cancel process group transition 
[2026-08-21 09:02:26.985] [INFO] [launch_manager] 2026/08/21 09:02:26.2946984 2386734 000 ECU1 LM LM log debug verbose 1 Control Client handler nudged 
[2026-08-21 09:02:26.985] [INFO] [launch_manager] 2026/08/21 09:02:26.2946984 2386735 000 ECU1 LM LM log debug verbose 1 Wait for process group cancellation
[2026-08-21 09:02:29.058] [INFO] [launch_manager] 2026/08/21 09:02:29.2949053 2407424 000 ECU1 LM LM log error verbose 1 NOTE: Cancellation timed out 
[2026-08-21 09:02:29.058] [INFO] [launch_manager] 2026/08/21 09:02:29.2949053 2407425 000 ECU1 LM LM log debug verbose 1 Start transitioning process group to Off state

So between the end of the test...
[2026-08-21 09:02:23.619] [INFO] [launch_manager] [  PASSED  ] 1 test.
...and the timed out cancellation...
[2026-08-21 09:02:26.985] [INFO] [launch_manager] 2026/08/21 09:02:26.2946984 2386735 000 ECU1 LM LM log debug verbose 1 Wait for process group cancellation
... already more then 3s are gone.

I am wondering if this error is related to the scheduling priorities of the application processes or if there is a real bug in launch manager.
As a workaround the timeout could be increased to 5s.

[MaciejKaszynski]

I also see 2s from cancellation to cancellation time out. I think our tests run on a single core machine so a scheduling prio thing could be the cause

[2026-08-21 11:24:19.882] [INFO] [launch_manager] 2026/08/21 11:24:19.1459881 156100098 000 ECU1 LM LM log debug verbose 1 Wait for process group cancellation
[2026-08-21 11:24:21.916] [INFO] [launch_manager] 2026/08/21 11:24:21.1461915 156120444 000 ECU1 LM LM log error verbose 1 NOTE: Cancellation timed out 

[TimoSteuerwaldETAS]

As @WilliamRoebuck is currently working SafeProcessMap, I will put this on hold.

[TimoSteuerwaldETAS]

Just wanted to add, that sandbox_options test seems to be stable in the pipeline on this draft pr:
https://github.com/eclipse-score/lifecycle/actions/runs/33496928778/job/99821150707?pr=575#step:5:454