async client behavour

[MrWaloo]

I have written something like the HA modbus integration but for another system called Jeedom. So users can configure it through a web interface. I have used the async clients.
When the user starts the daemon, the configuration is converted into ModbusRequests, then async client.connect() is called and the ModbusRequests are executed in a loop (the user can chose the polling time and other parameters, that's not the point).
The version 3.7.2 is used under python 3.11.
Anyway a user has reported that he gets some exceptions, always the same and only sometimes.:

Modbus Error: [Connection] Not connected[AsyncModbusTcpClient 192.168.1.24:502]

The exceptions are catched and logged.

client.close() is never called in the loop, I can guaranty that.

Can this exception occur because of the connection quality? I did not encounter such errors at home where all my Ethernet modbus equipment are wired.

[janiversen]

It means that the device is not connected to, and did not accept a connect within the configured retries.

I cannot tell you more without a debug log.

[MrWaloo]

The log may not be in the desired format... But it is the format of Jeedom. I have deleted the rows from my plugin but the request execute and the error.
The connection name is the Jeedom equipment name = "Wago" in this case.

So here is the log arround the error:

[2024-09-17 13:50:31][DEBUG] : Adding transaction 422
[2024-09-17 13:50:31][DEBUG] : Resetting frame - Current Frame in buffer - 
[2024-09-17 13:50:31][DEBUG] : send: 0x1 0xa6 0x0 0x0 0x0 0x6 0x1 0x1 0x0 0x9 0x0 0x1
[2024-09-17 13:50:31][DEBUG] : recv: 0x1 0xa6 0x0 0x0 0x0 0x4 0x1 0x1 0x1 0x0 old_data:  addr=None
[2024-09-17 13:50:31][DEBUG] : Processing: 0x1 0xa6 0x0 0x0 0x0 0x4 0x1 0x1 0x1 0x0
[2024-09-17 13:50:31][DEBUG] : Factory Response[ReadCoilsResponse': 1]
[2024-09-17 13:50:31][DEBUG] : Getting transaction 422
[2024-09-17 13:50:31][DEBUG] : -> transport: received eof
[2024-09-17 13:50:31][DEBUG] : Connection lost Wago due to None
[2024-09-17 13:50:31][DEBUG] : callback_disconnected called: None
[2024-09-17 13:50:31][DEBUG] : Wait Wago 100.0 ms before reconnecting.
[2024-09-17 13:50:31][DEBUG] : Wago: 'one_cycle_read'/Eclairage Ch Eliott ETAT: requesting read <<<<<---- At this point, the ModbusRequest has been executed
[2024-09-17 13:50:31][ERROR] : Wago/Eclairage Ch Eliott ETAT: exception during read request on slave id 1, address 2 -> Modbus Error: [Connection] Not connected[AsyncModbusTcpClient 192.168.1.24:502]
[2024-09-17 13:50:31][DEBUG] : Connecting Wago
[2024-09-17 13:50:31][DEBUG] : Connected to Wago
[2024-09-17 13:50:31][DEBUG] : Resetting frame - Current Frame in buffer - 
[2024-09-17 13:50:32][DEBUG] : Adding transaction 423
[2024-09-17 13:50:32][DEBUG] : Resetting frame - Current Frame in buffer - 
[2024-09-17 13:50:32][DEBUG] : send: 0x1 0xa7 0x0 0x0 0x0 0x6 0x1 0x1 0x0 0xd 0x0 0x1
[2024-09-17 13:50:32][DEBUG] : recv: 0x1 0xa7 0x0 0x0 0x0 0x4 0x1 0x1 0x1 0x0 old_data:  addr=None
[2024-09-17 13:50:32][DEBUG] : Processing: 0x1 0xa7 0x0 0x0 0x0 0x4 0x1 0x1 0x1 0x0
[2024-09-17 13:50:32][DEBUG] : Factory Response[ReadCoilsResponse': 1]
[2024-09-17 13:50:32][DEBUG] : Getting transaction 423

I guess, I could use the on_connect_callback callback to catch this kind of disconnection or just test client.connected since I use the async version of the client.

edit:
After having checked the code, it appears that the modbus slave closes the connection

-> transport: received eof

So I have implemented a on_connect_callback function that handels this case. Is that the proper way to do it?

[janiversen]

Your device or something in between closes the connect.
"-> transport: received eof"

Some devices needs a request every 20-30 seconds not to close the connection.