The RulesExecuted metric simply tracks the total number of IoT rule executions per region. Blue Matador detects anomalies in this metric which can be a leading indicator of performance issues. A sudden increase in rule executions can lead to increased strain on other topics, Kinesis streams, DynamoDB tables, or any service that your rule actions affect. A brief anomaly in this metric is usually not something to worry about and should only be used to correlate other issues in your system.
The ParseError metric is reported for each IoT Rule and keeps track of the number of JSON parse errors encountered. The likely cause of parse errors is a client sending malformed data. It is also possible that a Republish rule action is propagating malformed JSON. You will have to track down the source of the JSON errors in order to avoid parse errors.
The IoT Rules Engine keeps track of the number of messages processed and will throttle messages that go above an undocumented AWS-defined limit. It also detects malicious activity and will throttle messages that are suspected to be malicious. These events are recorded in the RuleMessageThrottled CloudWatch metric which Blue Matador monitors and will create a warning if any messages are throttled. If you encounter throttled messages, you should create a AWS support ticket to determine the reason for throttling.
An action failure occurs when a rule action fails to execute. This is usually because of a configuration issue, IAM permission error, or an issue with a 3rd-party service. Blue Matador monitors the Failure metric in CloudWatch for every combination of rule and action type. You may find it helpful to enable CloudWatch Logs for IoT to get more details about rule execution and determine the root cause of action failures.
The number of failures is reported to CloudWatch by the type of action, and there is no way to differentiate between multiple actions of the same type on the same rule. In order to make it easier to determine the cause of action failures, you can configure multiple rules instead of using more than one of the same action type on a rule.