How can I control the time out of waiting for response to an update publish

Question

I'm working with AFR and using Shadow and MQTT agent libs.
I have two named shadows.
I publish two update messages, one for each shadow, one after the other in a raw, and I see that the first shadow incoming publish update accepted message arrives after I send the second shadow update message and I get and error in prvIncomingPublishUpdateAcceptedCallback (That is based on the AFR demo shadow_device_task) saying that the ulClientToken is not the one the function is expected and thus it is ignored.
I can see that the ignored ulClientToken  is belong to the first update message and the comment in the function saying:
```
/* If we are waiting for a response, ulClientToken will be the token for the response
  * we are waiting for, else it will be 0. ulRecievedToken may not match if the response is
  * not for us or if it is is a response that arrived after we timed out
  * waiting for it.
*/
```
I guess my case is related to the last sentence in this comment and the response I got actually belong to the first update, so how I can control this time out that the comment is mention so I can enlarge it?
Thanks

Answer

Hi Eyal,

The problem is likely not a time out but the fact that the [demo code](https://github.dev/FreeRTOS/coreMQTT-Agent-Demos/blob/main/source/demo-tasks/shadow_device_task.c) is not designed to handle more than 1 in-flight update to shadow documents. The `ulClientToken` variable is global and can only store one value: if the second publish happens before the reception of the accepted/rejected messages for the first publish, its value gets overwritten. You could modify the sample to use a set/map/list to store all the pending client tokens and match them to the received client token in `prvIncomingPublishUpdateAcceptedCallback`.

Cheers,

Massimiliano

How can I control the time out of waiting for response to an update publish

関連するコンテンツ