Regression on nightly builds sros2.src.ros2.sros2.sros2.test.sros2.commands.security.verbs.test_generate_policy.test_generate_policy_services

[tfoote]

Bug report

sros2.src.ros2.sros2.sros2.test.sros2.commands.security.verbs.test_generate_policy.test_generate_policy_services is failing on multiple platforms on the nightly builds

Required Info:

• Operating System:
  • Linux aarch64r repeated: https://ci.ros2.org/job/nightly_linux-aarch64_repeated/1192/testReport/
  • Linux repeated: https://ci.ros2.org/job/nightly_linux_repeated/1871/testReport/
  • Linux release: https://ci.ros2.org/job/nightly_linux_release/1542/testReport/
  • Fci too: http://build.ros2.org/job/Fci__nightly-fastrtps_ubuntu_focal_amd64/120/testReport/
  • OSX debug: https://ci.ros2.org/job/nightly_osx_debug/1624/testReport/
  • OSX repeated: https://ci.ros2.org/job/nightly_osx_repeated/1953/testReport/
• Installation type:
  • CI builds
• Version or commit hash:
  • May 14th nightly
• DDS implementation:
  • Default
• Client library (if applicable):
  • N/A

Steps to reproduce issue

Run the CI:

sros2.src.ros2.sros2.sros2.test.sros2.commands.security.verbs.test_generate_policy.test_generate_policy_services

Expected behavior

Tests pass

Actual behavior

Tests fail

Additional information

[ruffsl]

The stacktrace from the failed test log doesn't seem informative:

test/sros2/commands/security/verbs/test_generate_policy.py:44: in test_generate_policy_topics
    assert cli.main(
E   AssertionError: assert 1 == 0
E    +  where 1 = <function main at 0xffffb1e529d0>(argv=['security', 'generate_policy', '/tmp/tmpi_bj77wj/test-policy.xml'])
E    +    where <function main at 0xffffb1e529d0> = cli.main

Is the python stacktrace being captured correctly?

[tfoote]

Well it's asserting here:

sros2/sros2/test/sros2/commands/security/verbs/test_generate_policy.py

Line 44 in 57e08d8

assert cli.main(

that the return code is zero so there's not much stack trace. The function has returned 1 already.

[kyrofa]

We've hit a difference in environment between the nightly tests and the normal ones in the past, could this be related to that?

[tfoote]

@kyrofa That would be surprising if something like that changed overnight, it's been working up until today.

@hidmic Pointed out to me that this could be related: #212

[sloretz]

If I understand correctly, @hidmic's PR #214 might fix this

[mikaelarguedas]

This should be addressed by #214 so I'm going to close this. Feel free to comment here if the issue persists and we can reopen