Health Monitoring of User Created Tasks

[ebenezerarunkumar]

For some components in our deployment, we need to create tasks using Os::Task within passive components rather than making the component active. For example, the Drv::LinuxUartDriver uses this pattern.

The Svc::Health component pings active components to detect hung tasks. However, for passive components with user-created tasks, there's no automatic health monitoring.

What is the recommended approach for health monitoring of user-created tasks in passive components?

Is it to make the component queued, add async Ping input ports, and then call doDispatch() within the custom task? Why does the task in Drv::LinuxUartDriver not ping the Svc::Health component?

[bocchino]

This is a very good question. Here is my take:

1. For doing work in response to port invocations, avoid using tasks inside passive components. Prefer using active components with health monitoring.
2. To block on external input, as in the UART driver, currently you have to use a task inside a passive component. We have an issue open to provide a better way of doing this. See Add a task component kind fpp#261. For now, a task inside a passive component is the way to do it.
3. When you are blocking on external input, you can't monitor health in the same way as for active components. The task is blocked on external input, so it can't respond to health pings. You could have some other kind of health monitoring. For example, you could have a command loss timer that resets the system if no input is received and processed for some amount of time.

[ebenezerarunkumar]

That makes sense! I think eventually having a task component kind defined in fpp is a good idea