Closed ahendrix closed 11 years ago
[gerkey] Not sure if this is related to #3984, so I didn't reopen that ticket.
[wim] I haven't been able to reproduce this on my machine (8 core, Karmic). What is your setup, and how often do you see this test fail?
The screens full of error messages are normal, those come from the 'stress test' of the controller manager.
[gerkey] The failure mentioned above is a single-core VM running 64-bit Jaunty. It doesn't seem to be failing on a 4-core physical machine running 32-bit Hardy (that's my other test setup).
If the test is load/time-dependent, you can disable it in CMake based on the number of available cores, or whether it's a VM:
[wim] The failure is very likely caused by high load (it is counting the number of messages received in a fixed time interval), so I disabled the test on a vm. Release of pr2_mechanism 1.0.4 will get pushed when build tests complete.
$ svn ci -m "only run test when not on vm because test is sensitive to load and timing" Sending pr2_controller_manager/CMakeLists.txt Transmitting file data . Committed revision r35824.
[wim] pr2_mechanism 1.0.4 released
[wim] This test is still failing.
[wim] Copied the boxturtle tag into the latest tag. Now latest will also not test on vm.
A test is consistently failing in pr2_controller_manager in boxturtle, e.g.:
http://build.willowgarage.com/view/ros-boxturtle/job/ros-boxturtle-all-test-ubuntu-jaunty-x86_64/2/testReport/pr2_controller_manager.TEST-test_controller_manager/TestController/pr2_controller_manager_test_controller_manager_publisher_hz/
I can reproduce this failure on my machine, though the exact numbers seem to vary. Also, I get screens full of red error messages.
trac data: