Open a sample report page

• Input:
  • CARET trace data (CTF file)
• Output (html pages):
  • Top page (index.html):
    • Links to each report page
    • Summary
  • Node analysis report:
    • Detailed information of each callback function
    • Verification results whether each timer/subscription callback function runs appropriately
  • Path analysis report:
    • Message flow graph and response time of each target path

• Ubuntu 20.04
• ROS2 Galactic
• CARET
• The followings

# Flask 2.1.0 (need to specify version) is required to create html report pages
pip3 install Flask==2.1.0

# firefox, selenium and geckodriver are required to generate graph image files
sudo apt install -y firefox
pip3 install selenium
wget https://github.com/mozilla/geckodriver/releases/download/v0.31.0/geckodriver-v0.31.0-linux64.tar.gz
tar xzvf geckodriver-v0.31.0-linux64.tar.gz
mv geckodriver /usr/local/bin/.

• It uses lots of memory
  • 64GB or more is recommended
  • In case crash happens due to memory shortage, increase swap space

Please refer to sample_autoware to find quick usage

• Run the following commands
  • Make sure to modify settings for your usage and environment
  • Make sure to prepare JSON files
• report_{dir_name_of_trace_data} is created
• Open index.html to see a report

# Settings: modify for your usage and environment
export script_path=./CARET_report                   # Path to this repo cloned
export package_list_json=./package_list.json        # Path to setting file you prepare
export target_path_json=./target_path.json          # Path to setting file you prepare
export trace_data=~/.ros/tracing/caret_sample/      # Path to CARET trace data (CTF file)
export start_time=0                                 # start time[sec] for analysis
export duration_time=9999                           # duration time[sec] for analysis
export max_node_depth=20                            # The number of depth to search path
export draw_all_message_flow=false                  # Flag to a create message flow graph for a whole time period (this will increase report creation time)

# Run script
sh ${script_path}/make_report.sh

• Node analysis report will be created for each package_name
• Settings
  • package_list_json : path to the JSON file
• Please describe the following information
  • Package name information (package_dict)
    • Pairs of package_name and regular expression for nodes belonging to the package
  • Ignore node list (ignore_list)
    • List of Regular expression for nodes to be ignored

{
    "package_dict": {
        "Package_A" : "^/package_a",
        "Package_B" : "^/package_b",
    },
    "ignore_list": [
        "noisy_node_name_a",
        "^/annoying_package_name",
    ]
}

• Path analysis report will show results for pathes described in this JSON file
• Settings
  • target_path_json : path to the JSON file
  • max_node_depth : In case a target path is not found, increase the number. In case the script stuck, descrease the number
• Please describe the following information
  • Pair of name and path
  • path is a list of node_name (note: path doean't mean path as filesystem!)
    • Regular expression is supported
    • You can also set [node_name, topic_name] instead of node_name . It is useful when two nodes are connected via multiple topics

[
    {
        "name": "path_name",
        "path": [
            "The first node_name in the path",
            "The second node_name in the path",
            ["The third node_name in the path", "The topic name published by the third node"]
            "...",
            "The last node_name in the path",
        ]
    }
    {
        "name": "sample_path",
        "path": [
            "/package_a/node_1",
            "/package_a/node_2",
            ["/package_a/node_3", "/topic_3"],
            "/package_a/node_4_.*"
        ]
    },
]

• Install Dear RosNodeViewer
• Open graph ( e.g.: dear_ros_node_viewer architecture.yaml )
• Select nodes in the path which you want to analyze
  • ctrl + click
• Press C to export node name list to clip board
• Paste the exported node name list to json file

• In case a node name varies for each execution, you can write node name in JSON like the following
  • Before: /node_name_xyz_abc1234567_1234567_1234567891234567891
  • After: /node_name_xyz.*