AMRNAV-6776 AMR diagnostics: Use CPU and RAM monitor from diagnostics #3

VladyslavHrynchak200204 · 2024-10-15T11:59:37Z

Reason for change:

https://github.com/ros/diagnostics/ now has a CPU and RAM monitor built-in. Please

create a new branch in our fork and update it to latest ros2 from upstream (contact me in case of doubts about conflicts)
replace the CPU and RAM monitor we currently use from system_monitor with the ones from diagnostics
compare their statistics on an AMR (ensure that they give similar values, if not, document how they change).
remove system_monitor from AMR.

Changes:

Added compatibility for CPU and RAM monitor from upstream
Changes for https://github.com/logivations/deep_cv/pull/8482

Result:

… marked discard_stale, or when all of their items are STALE. (ros#315) * If discard_stale = true, don't interpret stale as ERROR when rolling up * Check in test to exercise behavior --------- Signed-off-by: Andrew Symington <andrew.c.symington@gmail.com> Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com> Co-authored-by: Christian Henkel <christian.henkel2@de.bosch.com>

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

* add cpu monitor to the common diagnostics * add 3 unittests * add psutil to package xml * add debug print for cpu percentages and change checking percentage to -1 * update code with review comments. Added a SetUp() method in the test and move the sleep to there to make sure this is done every test in a systemactic way --------- Co-authored-by: Richardvdketterij <richard.van.de.ketterij@nobleo.nl>

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

This reverts commit 5077016.

* runs on container * all on ubuntu latest * pydocstyle fix is obsolete * no uncrustify for noble --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

- using a launchtest --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

* testing with ubuntu ntp server * message makes a little more sense with abs * bigger offsets allowed * organizing imports with isort profile `google` --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

* NTP monitor improvements * commit to run tests again * bring back double callback * remove whiteline --------- Co-authored-by: Angsa Deployment Team <team@angsa-robotics.com>

* refactor(ram_monitor): ros2 port * Update diagnostic_common_diagnostics/diagnostic_common_diagnostics/ram_monitor.py * Apply suggestions from code review * Update diagnostic_common_diagnostics/README.md * fstrings * python3 --------- Co-authored-by: Christian Henkel <6976069+ct2034@users.noreply.github.com>

* Fix usage of rclcpp::ok with a non-default context The current implementation calls `rclcpp::ok` without any arguments, which amounts to verifying that the global default context is valid. In the case where a node is added to a custom context, and the global context is not used, `rclcpp::ok` without any arguments will always return `false` since the global context has never been initialized. To fix it, pass to rclcpp the context that's available via the node's base interface. * Add a test for custom context

…egradation (ros#324) * Aggregator: publish diagnostics_toplevel_state immediately on every degradation * Update docs * Re-use the publishData function such that also the actual diagnostics are available * Expand test with more degradation cases

* Add add_analyzer functionality * Add copyright notice and license, remove unused includes, re-order includes correctly * Increase clarity of prefix_ by renaming it to analyzers_ns_ * Add add_analyzer functionality * Fix bug where base_path is not reset correctly * Make the parameter forwarding condition more generic, fix the default service namespace from diagnostics_agg to analyzers * Add an add_analyzer example to the diagnostic_aggregator * Update the relevant READMEs * Fix linter errors * Add test for add_analyzer at runtime, remove unnecessary ros info logger, remove unnecessary hardcoded namespace from yaml files * Remove the now redundant analyzers_ns_ * Change the copyright of add_analyzer, forgot to update it to Nobleo after copying the notice

thanks to https://github.com/reinzor Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

A new clock was created for every diagnosed publisher. When running with 'use_sim_time', the /clock time was not respected.

* Adoptin Ci changes similar to jazzy Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com> * also ignoring pep257 here ... Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com> --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

* Port hd_monitor.py * Adapt documentation and CMakeList to new hd_monitor.py * Fix execution flag of hd_monitor * Add launch test for hd_monitor * Implement low and crit parameters in hd_monitor * Improve hd_monitor code quality

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

…ix usage of rclcpp::ok with a non-default context (ros#352) (ros#390) * Minimize header includes by moving impl to .cpp files (ros#331) * Minimize header includes by moving impl to .cpp files * Make sure to build a shared library Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com> * Fix usage of rclcpp::ok with a non-default context (ros#352) * Fix usage of rclcpp::ok with a non-default context The current implementation calls `rclcpp::ok` without any arguments, which amounts to verifying that the global default context is valid. In the case where a node is added to a custom context, and the global context is not used, `rclcpp::ok` without any arguments will always return `false` since the global context has never been initialized. To fix it, pass to rclcpp the context that's available via the node's base interface. * Add a test for custom context Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com> --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com> Co-authored-by: Ramon Wijnands <ramon.wijnands007@gmail.com> Co-authored-by: Hervé Audren <101862279+haudren-woven@users.noreply.github.com>

Without this, downstream packages have missing symbols Relates to: ros#387 (cherry picked from commit d8fef69) Co-authored-by: Ramon Wijnands <ramon.wijnands@nobleo.nl>

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

…R-diagnostics-Use-CPU-and-RAM-monitor-from-diagnostics

VladyslavHrynchak200204 · 2024-10-15T14:12:57Z

diagnostic_common_diagnostics/setup.py

+    tests_require=["pytest"],
+    entry_points={
+        'console_scripts': [
+            'cpu_monitor = diagnostic_common_diagnostics.cpu_monitor:main', 


added setup.py to diagnostics_common_diagnostics package to easily call cpu_monitor and ram_monitor as executable

VladyslavHrynchak200204 · 2024-10-16T07:02:06Z

diagnostic_aggregator/CMakeLists.txt

@@ -72,6 +73,10 @@ target_link_libraries(aggregator_node

 rclcpp_components_register_nodes(${PROJECT_NAME} "diagnostic_aggregator::Aggregator")

+# Add analyzer
+add_executable(add_analyzer src/add_analyzer.cpp)


Resolved conflicts: added analyzer to diagnostic_aggregator/CMakeLists.txt

VladyslavHrynchak200204 · 2024-10-16T07:09:14Z

diagnostic_aggregator/src/aggregator.cpp

Resolve conflicts: added rclcpp::NodeOptions options to Aggregator::Aggregator constructor

VladyslavHrynchak200204 · 2024-10-16T07:28:11Z

diagnostic_aggregator/src/aggregator.cpp

+  {  // lock the mutex while analyzer_group_ and other_analyzer_ are being updated
+    std::lock_guard<std::mutex> lock(mutex_);
+    n_ = std::shared_ptr<rclcpp::Node>(this, [](rclcpp::Node *) {});
+    analyzer_group_ = std::make_unique<AnalyzerGroup>();


Resolve conflicts: added lock_guard

VladyslavHrynchak200204 · 2024-10-16T12:53:14Z

diagnostic_common_diagnostics/diagnostic_common_diagnostics/cpu_monitor.py

    # Create the node
    hostname = socket.gethostname()
-    node = Node(f'cpu_monitor_{hostname.replace("-", "_")}')
+    node = Node('cpu_monitor')

    # Declare and get parameters
    node.declare_parameter('warning_percentage', 90)


added get_cpu_diagnostics_node for easy launch from python_launch.py and changed node name to cpu_monitor instead of cpu_monitor_{hostname.replace("-", "_") not to be depended of hostname in yaml file

VladyslavHrynchak200204 · 2024-10-16T12:54:26Z

diagnostic_common_diagnostics/diagnostic_common_diagnostics/ram_monitor.py

-    rclpy.init()
-    node = rclpy.create_node(f'ram_monitor_{hostname.replace("-", "_")}')
+
+    node = rclpy.create_node('ram_monitor')


the same changed for ram_monitor as listed above

ct2034 and others added 30 commits January 18, 2024 23:03

formatting fixes from PR324 (ros#327)

ac97a0c

typo

183b7de

including depdency (ros#322)

a09c0b1

documentation on new branch

b1c8dd1

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Rolling obviously also uses the jazzy branch

83c030e

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

changelogs

07ed478

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

3.2.0

c1b908c

trying fix

5077016

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Revert "trying fix"

194753a

This reverts commit 5077016.

Building in docker (ros#335)

8ee5640

* runs on container * all on ubuntu latest * pydocstyle fix is obsolete * no uncrustify for noble --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Fixing ntp launchtest (ros#330)

431926f

- using a launchtest --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

only one distro per branch

1d19a20

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

readme update

e319448

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Using ubuntu ntp server in systemtest (ros#346)

c4f7231

* testing with ubuntu ntp server * message makes a little more sense with abs * bigger offsets allowed * organizing imports with isort profile `google` --------- Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

NTP monitor improvements (ros#342)

7efb71a

* NTP monitor improvements * commit to run tests again * bring back double callback * remove whiteline --------- Co-authored-by: Angsa Deployment Team <team@angsa-robotics.com>

refactor(sensors_monitor): ros2 port ros#339

e67a69c

thanks to https://github.com/reinzor Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

change(diagnosed-publisher): allow specifying node clock (ros#340)

16b5e8a

A new clock was created for every diagnosed publisher. When running with 'use_sim_time', the /clock time was not respected.

changelogs upated

8b0fb51

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

3.2.1

ac7cac9

4.3.0

97c65e3

versioning info (ros#373)

4f5ae69

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

writing down the backport tool and its usage (ros#377)

0af0de1

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Port hd_monitor to ROS2 (ros#334)

05a9645

* Port hd_monitor.py * Adapt documentation and CMakeList to new hd_monitor.py * Fix execution flag of hd_monitor * Add launch test for hd_monitor * Implement low and crit parameters in hd_monitor * Improve hd_monitor code quality

ct2034 and others added 8 commits July 17, 2024 14:51

fixing pep257 problems introduced by ros#334 (ros#384)

b504059

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Fix correctly exporting the library (ros#388) (ros#393)

275dd5e

Without this, downstream packages have missing symbols Relates to: ros#387 (cherry picked from commit d8fef69) Co-authored-by: Ramon Wijnands <ramon.wijnands@nobleo.nl>

changelogs

a223af1

4.3.1

cd24251

adding buildfarm statuses (ros#394)

bb392ee

Signed-off-by: Christian Henkel <christian.henkel2@de.bosch.com>

Merge remote-tracking branch 'upstream/ros2' into ros2-AMRNAV-6776-AM…

a2df5ba

…R-diagnostics-Use-CPU-and-RAM-monitor-from-diagnostics

Add setup.py for diagnostic_common_diagnostics package

59b45f6

VladyslavHrynchak200204 requested a review from jplapp October 15, 2024 12:00

VladyslavHrynchak200204 added task do not merge labels Oct 15, 2024

VladyslavHrynchak200204 commented Oct 15, 2024

View reviewed changes

VladyslavHrynchak200204 commented Oct 16, 2024

View reviewed changes

resolve conflicts: added lock_guard to aggregator.cpp

f392c12

VladyslavHrynchak200204 commented Oct 16, 2024

View reviewed changes

added function to get node and removed hostname from node name

3be0553

VladyslavHrynchak200204 commented Oct 16, 2024

View reviewed changes

jplapp approved these changes Oct 17, 2024

View reviewed changes

VladyslavHrynchak200204 added 4 commits October 21, 2024 10:55

fix typo in description

78ac516

fix errors

abbf2cc

fix errors

78af864

fix errors with whitespace

bbb18d0

VladyslavHrynchak200204 merged commit bbb18d0 into ros2 Oct 22, 2024
8 of 9 checks passed

VladyslavHrynchak200204 added do not delete and removed do not merge labels Oct 22, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

AMRNAV-6776 AMR diagnostics: Use CPU and RAM monitor from diagnostics #3

AMRNAV-6776 AMR diagnostics: Use CPU and RAM monitor from diagnostics #3

VladyslavHrynchak200204 commented Oct 15, 2024 •

edited

Loading

VladyslavHrynchak200204 Oct 15, 2024

VladyslavHrynchak200204 Oct 16, 2024

VladyslavHrynchak200204 Oct 16, 2024

VladyslavHrynchak200204 Oct 16, 2024

VladyslavHrynchak200204 Oct 16, 2024

VladyslavHrynchak200204 Oct 16, 2024

AMRNAV-6776 AMR diagnostics: Use CPU and RAM monitor from diagnostics #3

AMRNAV-6776 AMR diagnostics: Use CPU and RAM monitor from diagnostics #3

Conversation

VladyslavHrynchak200204 commented Oct 15, 2024 • edited Loading

VladyslavHrynchak200204 Oct 15, 2024

Choose a reason for hiding this comment

VladyslavHrynchak200204 Oct 16, 2024

Choose a reason for hiding this comment

VladyslavHrynchak200204 Oct 16, 2024

Choose a reason for hiding this comment

VladyslavHrynchak200204 Oct 16, 2024

Choose a reason for hiding this comment

VladyslavHrynchak200204 Oct 16, 2024

Choose a reason for hiding this comment

VladyslavHrynchak200204 Oct 16, 2024

Choose a reason for hiding this comment

VladyslavHrynchak200204 commented Oct 15, 2024 •

edited

Loading