Fix test_security tests using rmw_connextdds by disabling Monitoring 2.0

[fgallegosalido]

Description

With the enabling of the Monitoring 2.0 Library in rolling (see ros2/rmw_connextdds#243), security tests are failing because the Monitoring Participant must be secured if other Participants are also secured.

To fix these tests, we explicitly disable Monitoring 2.0 through the environment variable RTI_MONITORING2_ENABLE.

Is this user-facing behavior change?

Tests in test_security using the rmw_connextdds will now pass again.

Did you use Generative AI?

No

Additional Information

Monitoring is specific to the Connext RMW. We don't need to test for it in system_tests.

[mergify]

Tick the box to add this pull request to the merge queue (same as @mergifyio queue).

• Queue this pull request

[fgallegosalido]

@fujitatomoya can you launch a validation?

[jmachowinski]

@fgallegosalido can you fix the DCO ?

[jmachowinski]

Pulls: #596
Gist: https://gist.githubusercontent.com/jmachowinski/6525b134911182ee8bf63e33af6e9635/raw/13437f4a0c796af0eca8b8f59e5b918d33cad8fa/ros2.repos
BUILD args:
TEST args:
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/19707

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[fgallegosalido]

@jmachowinski Sorry, I forgot to sign off the commit. I will wait for the CI to pass and open a new pr with the commit correctly signed. Is there any other way of doing it?

In any case, I'll try doing what the action says and see how it goes

[fgallegosalido]

@jmachowinski I think it is fixed. Let me know if this works.

[jmachowinski]

@fgallegosalido

Looks good, now we just wait for CI ;-)

As the commit is identical the CI job is valid in any case.

[fgallegosalido]

@jmachowinski @fujitatomoya the Linux failure is not related to the Connext RMW. The Windows build is interesting, as it seems to have marked as failures tests that passed. Is that expected?

[jmachowinski]

No, this looks like a infrastructure issue, I'll ping the infra team.

[jmachowinski]

I just realized I ran the CI with the wrong parameters. Almost all test ran using FastDDS...

[jmachowinski]

Pulls: #596
Gist: https://gist.githubusercontent.com/jmachowinski/0f922e6ee0631440fa36815351a94ee9/raw/13437f4a0c796af0eca8b8f59e5b918d33cad8fa/ros2.repos
BUILD args: --cmake-args -DRMW_IMPLEMENTATION=rmw_connextdds
TEST args:
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/19723

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[fgallegosalido]

@jmachowinski the package gz_math_vendor couldn't be built. It seems the machine run out of memory.

[jmachowinski]

Pulls: #596
Gist: https://gist.githubusercontent.com/jmachowinski/037fc5705b6ff0c60509f37539cf2604/raw/13437f4a0c796af0eca8b8f59e5b918d33cad8fa/ros2.repos
BUILD args: --cmake-args -DRMW_IMPLEMENTATION=rmw_connextdds
TEST args:
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/19734

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[fgallegosalido]

@jmachowinski it seems some tests have been marked as failed, although they passed. The few tests that failed don't seem related to the Connext RMW. Also, I don't see the security tests run correctly. Only the default tests are run:

100% tests passed, 0 tests failed out of 6

Label Time Summary:
copyright     =   0.34 sec*proc (1 test)
cppcheck      =   0.27 sec*proc (1 test)
cpplint       =   0.36 sec*proc (1 test)
lint_cmake    =   0.27 sec*proc (1 test)
linter        =   1.89 sec*proc (6 tests)
uncrustify    =   0.30 sec*proc (1 test)
xmllint       =   0.34 sec*proc (1 test)

Total Test time (real) =   1.89 sec
---
Finished <<< test_security [2.42s]

I would expect more tests to be run. Can you look into it? Also, can we run just the test_security package? To make sure the changes in this pull request are correct.

[jmachowinski]

@fgallegosalido Is it correct that connect is not available for rhel and aarch64 ?

Are the security test run locally if you do colcon test ?

The CI does nothing else in the end.

If I run the CI starting from a pkg XYZ, the CI pulls binary packages from rolling for an pkg before XYZ,
AFAIK are the binary pkgs not up to date at the moment, as the infra team is overhauling the concept for them.

[fgallegosalido]

It's correct that rhel and aarch64 are not working at the moment (although we are working to get aarch64 to run. See ros2/ci#882).

I can run the security tests locally without any issues:

82% tests passed, 18 tests failed out of 99

Label Time Summary:
copyright      =   0.20 sec*proc (1 test)
cppcheck       =   0.18 sec*proc (1 test)
cpplint        =   0.23 sec*proc (1 test)
gtest          =   0.67 sec*proc (4 tests)
launch_test    = 551.05 sec*proc (88 tests)
lint_cmake     =   0.17 sec*proc (1 test)
linter         =   1.99 sec*proc (6 tests)
uncrustify     =   0.19 sec*proc (1 test)
xmllint        =   1.01 sec*proc (1 test)

Total Test time (real) = 554.15 sec

The following tests FAILED:
	 26 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__non_secure_comm_0 (Timeout) launch_test
	 27 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__non_secure_comm_1 (Timeout) launch_test
	 28 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__non_secure_comm_2 (Timeout) launch_test
	 29 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__non_secure_comm_3 (Timeout) launch_test
	 30 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__non_secure_comm_4 (Timeout) launch_test
	 31 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__secure_comm_0 (Timeout) launch_test
	 32 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__secure_comm_1 (Timeout) launch_test
	 33 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__secure_comm_2 (Timeout) launch_test
	 34 - test_secure_publisher_subscriber__Empty__rmw_cyclonedds_cpp__secure_comm_3 (Timeout) launch_test
	 37 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__non_secure_comm_0 (Timeout) launch_test
	 38 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__non_secure_comm_1 (Timeout) launch_test
	 39 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__non_secure_comm_2 (Timeout) launch_test
	 40 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__non_secure_comm_3 (Timeout) launch_test
	 41 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__non_secure_comm_4 (Timeout) launch_test
	 42 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__secure_comm_0 (Timeout) launch_test
	 43 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__secure_comm_1 (Timeout) launch_test
	 44 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__secure_comm_2 (Timeout) launch_test
	 45 - test_secure_publisher_subscriber__UnboundedSequences__rmw_cyclonedds_cpp__secure_comm_3 (Timeout) launch_test
Finished <<< test_security [9min 15s]   [ with test failures ]

Summary: 1 package finished [9min 16s]
  1 package had test failures: test_security

The failing tests are not related to the Connext RMW (even before the changes in this pr).