Lost connection to inverter after changing settings multiple times in quick succession, needs HA restart

[canton7]

From #40 (reply in thread):

The only issue was the one I mentioned with going in/out of the config repeatedly to change poll settings - after 3 or 4 times in short succession it goes into error (see below), I tried deleting the integration and adding again but it failed as it couldn't access the IP address - it needed an HA restart to resolve.

These were the logs when it lost it:-

2023-04-25 12:04:08.819 DEBUG (SyncWorker_6) [custom_components.foxess_modbus.modbus_client] Connecting to ModbusBaseClient._params(host='192.168.0.238', port=502, framer=<class 'pymodbus.framer.socket_framer.ModbusSocketFramer'>, timeout=3.0, retries=3, retry_on_empty=False, close_comm_on_error=False, strict=True, broadcast_enable=False, kwargs={'modbus_type': 'TCP'}, reconnect_delay=100, reconnect_delay_max=300000, baudrate=None, bytesize=None, parity=None, stopbits=None, handle_local_echo=None, source_address=None, sslctx=None, certfile=None, keyfile=None, password=None, server_hostname=None)
2023-04-25 12:04:08.824 DEBUG (MainThread) [custom_components.foxess_modbus.modbus_controller] Modbus exception when polling tcp://192.168.0.238:502 247 - Modbus Error: [Input/Output] Error reading registers from tcp://192.168.0.238:502: Modbus Error: [Input/Output] [Errno 104] Connection reset by peer
2023-04-25 12:04:14.817 DEBUG (MainThread) [custom_components.foxess_modbus.modbus_controller] Reading addresses on tcp://192.168.0.238:502 247: (31000, 25)
2023-04-25 12:04:14.820 DEBUG (SyncWorker_2) [custom_components.foxess_modbus.modbus_client] Connecting to ModbusBaseClient._params(host='192.168.0.238', port=502, framer=<class 'pymodbus.framer.socket_framer.ModbusSocketFramer'>, timeout=3.0, retries=3, retry_on_empty=False, close_comm_on_error=False, strict=True, broadcast_enable=False, kwargs={'modbus_type': 'TCP'}, reconnect_delay=100, reconnect_delay_max=300000, baudrate=None, bytesize=None, parity=None, stopbits=None, handle_local_echo=None, source_address=None, sslctx=None, certfile=None, keyfile=None, password=None, server_hostname=None)
2023-04-25 12:04:14.823 ERROR (SyncWorker_2) [pymodbus.logging] Connection to (192.168.0.238, 502) failed: [Errno 104] Connection reset by peer
2023-04-25 12:04:14.831 DEBUG (MainThread) [custom_components.foxess_modbus.modbus_controller] Modbus exception when polling tcp://192.168.0.238:502 247 - Modbus Error: [Connection] Failed to connect[ModbusTcpClient(192.168.0.238:502)]
2023-04-25 12:04:20.818 DEBUG (MainThread) [custom_components.foxess_modbus.modbus_controller] Reading addresses on tcp://192.168.0.238:502 247: (31000, 25)
2023-04-25 12:04:20.820 DEBUG (SyncWorker_0) [custom_components.foxess_modbus.modbus_client] Connecting to ModbusBaseClient._params(host='192.168.0.238', port=502, framer=<class 'pymodbus.framer.socket_framer.ModbusSocketFramer'>, timeout=3.0, retries=3, retry_on_empty=False, close_comm_on_error=False, strict=True, broadcast_enable=False, kwargs={'modbus_type': 'TCP'}, reconnect_delay=100, reconnect_delay_max=300000, baudrate=None, bytesize=None, parity=None, stopbits=None, handle_local_echo=None, source_address=None, sslctx=None, certfile=None, keyfile=None, password=None, server_hostname=None)
2023-04-25 12:04:20.823 ERROR (SyncWorker_0) [pymodbus.logging] Connection to (192.168.0.238, 502) failed: [Errno 104] Connection reset by peer
2023-04-25 12:04:20.827 DEBUG (MainThread) [custom_components.foxess_modbus.modbus_controller] Modbus exception when polling tcp://192.168.0.238:502 247 - Modbus Error: [Connection] Failed to connect[ModbusTcpClient(192.168.0.238:502)]

[canton7]

I think I might have seen this as well using UDP... I put it down to the W610 not liking connection from multiple sources at the same time, but it might actually be this (with a slightly different message of course, because UDP).

[canton7]

I've hit this a few times today, and I've had Wireshark open the entire time. Each time, the problem has been with the adapter -- we send a correct packet, but the adapter never responds.

[FozzieUK]

@canton7 yeah happened to me again last night, I only made a single change this time and it lost it, sounds like something in the underlying python libraries is getting messed up. For the average user it won’t happen very often and a restart fixes it so perhaps keep this on the back burner.

[canton7]

I'm not sure if it can be pymodbus -- the packets it's sending out look fine. It's just that (in my case) the adapter wasn't responding

[FozzieUK]

Ahh ok, maybe the socket isn't being reset properly after the change and the adaptor is left hung - possibly not much you can do about this one. Next time it happens rather than restart i'll go and pull the ethernet cable out of the adaptor and see if that helps it tidy up.

[canton7]

Well, I'm looking at UDP and the W610 -- there's no socket state there.

[canton7]

It's entirely possible we're seeing different issues, of course!

[canton7]

I noticed that the TCP connection to the inverter/adapter wasn't being properly closed when the integration was reloaded, which might be related: I can imagine lots of not-yet-timed-out TCP sockets causing the adapter/inverter some issues

[FozzieUK]

Ok, fingers crossed that's what it is - ironically last week I spent over 10 minutes going in / out of the config changing the number of reads, poll time etc.. and it didn't break once 😣 😏