AWS IoT Core throttling cool down?

Question

Hi,

we expect high traffic in our AWS IoT Core endpoint and performed load testing recently to understand its limits. After sending thousands of events it started to throttle and ignore our messages which is perfectly fine considering the limits defined [here](https://docs.aws.amazon.com/general/latest/gr/iot-core.html#limits_iot):

```
{
    "timestamp": "2022-02-04 15:55:46.100",
    "logLevel": "ERROR",
    "traceId": "-----------",
    "accountId": "-----------",
    "status": "Failure",
    "eventType": "Publish-In",
    "protocol": "MQTT",
    "topicName": "-------------",
    "clientId": "-------------",
    "principalId": "-------------",
    "sourceIp": "-------------",
    "sourcePort": 55497,
    "reason": "throttled"
}
```

The problem is it seems it never "cools down". If we give it one hour or days, the message is still rejected (Failure/throttled). 
I thought that maybe a new clientid could help it didn't.

Is this known and expected? If so, how can we deal with it? 
It's been hard to understand what is being taken into consideration for the throttling once it's active, maybe the sourceIp?

Answer

The "throttled" reason is quite common. How can we know the root cause of throttled? Do AWS have any monitor tools to figure out what the limit is reached?

Answer

Problem solved. The throttling was occurring because I had reached the limit of retained messages for the account.
If I delete them all or start sending messages with no retain flag it all goes well.
I wish the error message was clearer about it but I can't blame anyone else.

AWS IoT Core throttling cool down?

Relevant content