Detector model restrictions and limitations - AWS IoT Events

Detector model restrictions and limitations

The following things are important to consider when creating a detector model.

How to use the actions field

The actions field is a list of objects. You can have more than one object, but only one action is allowed in each object.

Example
"actions": [ { "setVariable": { "variableName": "pressureThresholdBreached", "value": "$variable.pressureThresholdBreached - 1" } } { "setVariable": { "variableName": "temperatureIsTooHigh", "value": "$variable.temperatureIsTooHigh - 1" } } ]
How to use the condition field

The condition is required for transitionEvents and is optional in other cases.

If the condition field isn't present, it's equivalent to "condition": true.

The result of the evaluation of a condition expression should be a Boolean value. If the result isn't a Boolean value, it's equivalent to false and won't initiate the actions or transition to the nextState specified in the event.

Availability of variable values

By default, if the value of a variable is set in an event, its new value isn't available or used to evaluate conditions in other events in the same group. The new value isn't available or used in an event condition in the same onInput, onEnter or onExit field.

Set the evaluationMethod parameter in the detector model definition to change this behavior. When the evaluationMethod is set to SERIAL, variables are updated and event conditions are evaluated in the order that the events are defined. Otherwise, when the evaluationMethod is set to BATCH or defaults to it, variables within a state are updated and events within a state are performed only after all event conditions are evaluated.

In the "Dangerous" state, in the onInput field, "$variable.pressureThresholdBreached" is decremented by one in the "Pressure Okay" event when the condition is met (when the current input has pressure less than or equal to 70).

{ "eventName": "Pressure Okay", "condition": "$input.PressureInput.sensorData.pressure <= 70", "actions": [ { "setVariable": { "variableName": "pressureThresholdBreached", "value": "$variable.pressureThresholdBreached - 1" } } ] }

The detector should transition back to the "Normal" state when "$variable.pressureThresholdBreached" reaches 0 (that is, when the detector has received three contiguous pressure readings less than or equal to 70). The "BackToNormal" event in transitionEvents must test that "$variable.pressureThresholdBreached" is less than or equal to 1 (not 0), and also verify again that the current value given by "$input.PressureInput.sensorData.pressure" is less than or equal to 70.

"transitionEvents": [ { "eventName": "BackToNormal", "condition": "$input.PressureInput.sensorData.pressure <= 70 && $variable.pressureThresholdBreached <= 1", "nextState": "Normal" } ]

Otherwise, if the condition tests for only the value of the variable, two normal readings followed by an over-pressure reading would fulfill the condition and transition back to the "Normal" state. The condition is looking at the value that "$variable.pressureThresholdBreached" was given during the previous time an input was processed. The value of the variable is reset to 3 in the "Overpressurized" event, but remember that this new value is not yet available to any condition.

By default, every time a control enters the onInput field, a condition can only see the value of a variable as it was at the start of processing the input, before it's changed by any actions specified in onInput. The same is true for onEnter and onExit. Any change made to a variable when we enter or exit the state isn't available to other conditions specified in the same onEnter or onExit fields.

Latency when updating a detector model

If you update, delete, and recreate a detector model (see UpdateDetectorModel), there is some delay before all spawned detectors (instances) are deleted and the new model is used to recreate the detectors. They are recreated after the new detector model takes effect and new inputs arrive. During this time inputs might continue to be processed by the detectors spawned by the previous version of the detector model. During this period, you might continue to receive alerts defined by the previous detector model.

Spaces in input keys

Spaces are allowed in input keys, but references to the key must be enclosed in backticks, both in the definition of the input attribute and when the value of the key is referenced in an expression. For example, given a message payload like the following:

{ "motor id": "A32", "sensorData" { "motor pressure": 56, "motor temperature": 39 } }

Use the following to define the input.

{ "inputName": "PressureInput", "inputDescription": "Pressure readings from a motor", "inputDefinition": { "attributes": [ { "jsonPath": "sensorData.`motor pressure`" }, { "jsonPath": "`motor id`" } ] } }

In a conditional expression, you must refer to the value of any such key using backticks also.

$input.PressureInput.sensorData.`motor pressure`