This package contains an implementation of a "remote" watchdog. There exists one watchdog monitor that monitors all watchdogs and many watchdogs.
Implementing a Watchdog in your node
To implement a watchdog timer into your code, you should first add the watchdog_timer package as a dependency to your package.
Then you should include the
in your code.
In addition, at the CMakeLists.txt file you should link your executable to the Watchdog library
Creating a Watchdog
To create a watchdog object (and register it at the watchdog monitor), create a watchdog object
Where the first argument of the constructor is the unique name of the watchdog and the second argument is the amount of time that the timer will countdown, until it resets.
Starting, Stopping and Reseting a Watchdog
Having your watchdog timer you can start/ stop/ reset the timer by using the objects methods.
By calling the
you can start the watchdog.
When a watchdog times out
When a watchdog times-out the watchdog monitor will emit a ROS_ERROR and will publish a message at the /robot/watchdogResets topic.
Starting the Watchdog monitor
To start the watchdog monitor you can run
rosrun watchdog_timer wdtMonitor
or you could include it in a .launch file
This watchdog implementation does not try to reset any nodes or perform any other action. It's sole purpose is to notify that a node has failed to reset in time a watchdog.
The watchdog implementation has a couple of services and topics. At the /robot/watchdogRegistry service a watchdog can register/ start/ stop itself. This is taken care by the client watchdog implementation. The /robot/watchdog topic is used for the watchdog client to publish their resets, to be heard by the watchdog monitor.