Multiple Callback for GET_POST_Allowed object

Pelion Device Management [forum moved]
1

I have created my new object under 3336/0/0, with following method.

static M2MResource location_res = add_resource(&obj_list, 3336, 0, 0, “Location_resource”, M2MResourceInstance::STRING,M2MBase::GET_POST_ALLOWED, “”, false, callback(&get_location));*

I have set callback for get_location.

once i use post option for this object from cloud page, my callback is triggered for 5 times.

I was expecting it should be for 1 time only.

I have analysed, every 2 sec my callback is triggered for 5 times.

if my callback is finished within 2 sec, i am not getting multiple trigger.

is it the right behaviour ??

If yes, Is there is any way , so i can send response to cloud, so that it can not trigger again and again ??

2

Hi, could you confirm which SDK you are using? Thanks

3

Hi ispacedroid,

I’ve seen similar behavior to this before. I believe the callback needs to finish quickly, otherwise Mbed Cloud will reissue the POST message to the device, triggering your callback again.

I’d recommend having your callback only set a flag by using some of the RTOS synchronization features to trigger your longer-running analyze step in a separate thread. Here’s the docs page on the RTOS apis: https://os.mbed.com/docs/v5.7/reference/rtos.html

Hope that helps, let me know if you have further questions!

Thanks,
Brian

4

So what Brian has said is correct, the retry mechanism kicks in when your callback does not finish fast enough. You can either use a Semaphore to switch from the callback back to main thread and execute there, or use mbed-events to do so.

F.e.:

static M2MResource location_res = add_resource(&obj_list, 3336, 0, 0, 
    "Location_resource", M2MResourceInstance::STRING,M2MBase::GET_POST_ALLOWED, 
    "", false, 
    queue.event(&get_location)); // <-- this is eventqueue magic

Will execute get_location on the event loop rather than blocking the POST. This should resolve any issues you see.