rai_state_logs

Runtime log digests for LLM-driven debugging, summaries and state updates.

How does it work?

The node collects logs in the system based on severity and, optionally, a user-set filter, up to a certain line count, keeping a rolling window.

• By default logs with level >= WARN are collected.

Parameters

• with clear_on_retrieval set to false logs are cleared after each retrieval (service call).
• when filters param contains a list of strings, logs of any severity that include these sub-strings are collected. By default, filters are empty, which means logs are collected by severity (warning or higher) alone.
• max_lines - lets you set the maximum number of lines to be stored in the log. 512 by default. When this number is exceeded, oldest logs are removed from the rolling window.
• include_meta, - lets you include meta information in the log, which is its severity, timestamp and source code location. True by default.

See launchfile: ./launch/rai_state_logs.launch.py for the easiest way to set these parameters.

Running with ros2 launch

ros2 launch rai_state_logs rai_state_logs.launch.py

Using keywords for filters:

ros2 launch rai_state_logs rai_state_logs.launch.py filters:="[\"Keyword1\", \"Keyword2\"]"