On ems-esp restart / upgrade old mqtt commands are re-executed (too early)

[tp1de]

On restart of ems-esp gateway old mqtt commands are executed again.
The execution is directly after detection of devices when not all entities are still discoverd.
This leads to errors in log.

Expected behavior: On any re-start old commands should not be executed.

[proddy]

Something is publishing these MQTT commands to EMS-ESP. EMS-ESP just subscribes and listens, it doesn't call mqtt commands by itself. I think you have them as retained messages so when EMS-ESP starts and subscribes, they start coming in again.

[tp1de]

Something is publishing these MQTT commands to EMS-ESP. EMS-ESP just subscribes and listens, it doesn't call mqtt commands by itself. I think you have them as retained messages so when EMS-ESP starts and subscribes, they start coming in again.

I will check for retained.
But every message has a timestamp which I see with mqtt explorer. Isn't it possible to check the timestamp while restarting?

[tp1de]

The commands are issued from node-red workflow within ha. (my heat demand function)

[tp1de]

.... and yes it does not happen without retained flag.

[proddy]

ideally, you shouldn't persist in published messages. Otherwise, there's no point in having a queue.

Our MQTT client library is based on the 3.1.1 protocol and version 5.0 supports more subscription features like "Retain As Published" and "Retain Handling" which we need here.

This is not an issue in the s/w (it's been like this for 5 years) so will move this to a discussion while we explorer updating the client.