@st3vev thx so much. This should really be mentioned in the tutorial. I also think that this polling needs to be activated automatically once a subscribe happens.

@ltn22 your are calling this method not on a bare metal system. There is an operating system running in the background and on top of it runs python. Each of them needs some memory to run. You can try to disable all radio devices, background tasks and place the your script in main.py to get rid of different timings by typing. But even your chance to see the same amount of free memory is low. There will be always small differences in the state of your LoPys which can result in a different memory usage. How well the collect can clean up your unused memory depends on the implementation. You can try to call collect multiple times and wait between the calls. But each additonal (high level) statement may result in a different state of the whole system. If you have trouble to get free memory for storing large arrays, there will be another problem. Fragmentation. mem_free may show you enough memory, but most likely this memory will not be available in one huge chunk. There will be a mixture of small and huge blocks depending on what was running before and if the garbage collector is able to move still allocated objects inside the memory.

@xykon Ok, thanks. Downgrading LoPy did the trick, but I still find it very strange that a working setup stops working due to a code problem after +70 hours of normal behavior and no longer starts working even after factory resetting, flashing the latest firmware again and starting again with the exact python project code and firmware... So what should I do next? Wait for a new LoPy firmware when the problem with get_wake_status is resolved? Can I follow some GitHub repository or just follow the forum here on this?

Thanks Meirssv for posting a solution. Had the same problem. Funny thing is that TTN only repeats the downlink payload on my LoPy module and not on my Microchip RN2483 module. It's like a downlink handshake from LoPy is missing, so TTN repeats queueing the payload. Can't find any documentation on downlink acknowledgements though. Anyway it works now thanks to your reply.

@jfs yes, shall be OK. Simply give it a try.

@jmarcelino That worked :D, was so simple... Thankyou

For anybody else that might have this problem, before trying to erase the flash you need to have a wire connected between G23 and GND and reset the board.

@seb i trying to run the code is Eexample's code LoRaWAN Nano-Gateway but i updating firmware hava solve it, :)

@seb bolded text Thank you your valuable information

@livius Thanks a lot. Now both of my LoPy works in AP mode. com port issue is adressed and fix will be released in next firmware @daniel can say when. I hope it is very soon.

I'm glad you resolved your problems. I'll post back in this thread when I know the issue with the newer firmware is resolved.

@crumble You are correct the code gets the time from a NTP server via WiFi but the example does not connect to any WiFi networks with internet access so this fails hence the time remains at 1970

I think you are right. I don't know how it was stuck in my mind that I could not change the join DR. Maybe by older topics of the forum. Thank you!

@dylan Instant connection, I left my Gateway on The Things Network as the meshed router, but changed the server in my config.py to router.au.thethings.network. Thank you :).

@livius OMG! it works! i tried s.send('ADC value:' + str(value)) but i never try s.send("ADC value:" + str(value)) thank you so much!