Socket leak when integration is started with Dyson powered off about ha-dyson HOT 6 CLOSED

Thanks for the information! I'll look into it and try to find a fix.

from ha-dyson.

I released version 0.9.1 that tries to fix this. Can you test that? Currently I don't have access to any Dyson device to test with.

from ha-dyson.

I'll give it a try, thanks!!

from ha-dyson.

Unfortunately the issue still persists. I'll try to debug it further.

from ha-dyson.

Just to add a bit more: calling disconnect on the libdyson device object disconnects the MQTT client and stops the loop, but the MQTT client doesn't clean up its local sockets in that case. This only seems to happen on reconnect or when the paho.mqtt.client.Client object is destroyed or reinitialised (https://github.com/eclipse/paho.mqtt.python/blob/master/src/paho/mqtt/client.py#L660). So something is holding on to a reference to the client object (presumably via the libdyson device object) even when the connection fails.

I haven't had time to debug further, but ideally these objects would not be referenced after the failed connection attempt and get cleaned up. If that can't be fixed, the libdyson device disconnect function could drop its reference to the MQTT client object, or reinitialise it.

from ha-dyson.

It just occurred to me that there is likely a cyclic dependency between the MQTT client and the libdyson device, perhaps that is getting in the way of the __del__ function being called.

from ha-dyson.