Fix system_tests flaky issue

[AlexeyMerzlyakov]

Testing PR for reproduce Issue No.1 from ticket #1585

[SteveMacenski]

Rebase on master. Those have been fixed

[AlexeyMerzlyakov]

Rebase on master. Those have been fixed

I've intentionally cherry-picked this commit from master in order to try to reproduce the problem on old branch. On latest master I've already tried unsuccessfully. Currently, I see this flaky problem (timeout on test_planner_costmaps and test_planner_random) looks like disappeared in CI.

[SteveMacenski]

You intentionally want all the cppcheck issues and map server not returning a result?

As far as I can tell, its disppeared in CI because you're not running it. I don't see in this PR where you uncommented the planner tests from the nav2_system_tests CMakeLists

[AlexeyMerzlyakov]

In this testing PR there is a lot of mess. I've tried both cases - to move test-issue1 branch over latest master with enabling system testing back (commit ID: e2e8e70) and moving test-issue1 branch over old reference commit where timeout problem was actually detected. Here was required some back-port from master which you've seen previously. I've checked - both cases did not give me a positive result: testing of test_planner_random passes w/o timeouts we saw previously.

Returning back test-issue1 branch to be over latest master to be consistent.

However, I've detected (and looks like fixed) new problem No.4 concerned with test_planner_costmaps failure on release testing only in Focal Fossa. I will describe it in the #1585 ticket since it will require some actions, I think.

[SteveMacenski]

OK, @AlexeyMerzlyakov you still have failures for linting, I think you just need to pull in changes / rebase, its been dealt with?

I would like to have this test run 5x in a row with no failures (except for the action utils, which is another subject altogether). It needs to pass at least 1 time all the way through to generate the new coverage report, so that rebase is required.

I ran it 5x already for you without an issue.

I’m ok to merge this when we can get a coverage report.

[codecov]

Codecov Report

Merging #1723 into master will increase coverage by 0.53%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1723      +/-   ##
==========================================
+ Coverage   63.52%   64.06%   +0.53%     
==========================================
  Files         191      192       +1     
  Lines       10300    10301       +1     
==========================================
+ Hits         6543     6599      +56     
+ Misses       3757     3702      -55     

Flag	Coverage Δ
#project	64.06% <ø> (+0.53%)	⬆️

Impacted Files	Coverage Δ
nav2_recoveries/plugins/wait.cpp	0.00% <0.00%> (-100.00%)	⬇️
nav2_costmap_2d/plugins/voxel_layer.cpp	0.00% <0.00%> (-44.30%)	⬇️
...costmap_2d/include/nav2_costmap_2d/voxel_layer.hpp	0.00% <0.00%> (-12.00%)	⬇️
..._voxel_grid/include/nav2_voxel_grid/voxel_grid.hpp	60.62% <0.00%> (-0.79%)	⬇️
nav2_util/include/nav2_util/costmap.hpp	100.00% <0.00%> (ø)
nav2_controller/src/nav2_controller.cpp	78.64% <0.00%> (+0.52%)	⬆️
nav2_navfn_planner/src/navfn.cpp	81.50% <0.00%> (+0.64%)	⬆️
nav2_costmap_2d/src/costmap_2d_publisher.cpp	45.16% <0.00%> (+0.80%)	⬆️
nav2_costmap_2d/plugins/static_layer.cpp	63.03% <0.00%> (+1.21%)	⬆️
nav2_costmap_2d/plugins/inflation_layer.cpp	97.90% <0.00%> (+1.39%)	⬆️
... and 5 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 433c6ac...beac064. Read the comment docs.