Can't connect to Elfin EW11a

[mschae]

Version of the custom_component

v1.5.2

Configuration

Add your logs here.

Describe the bug

When attempting to connect to the adapter for the first time I get the following error message:

Unable to connect to your adapter. Ensure that the adapter is properly configured (see the setup link above) and the connection details are correct, then try again. Details: Modbus Error: [Connection] ModbusTcpClient(10.112.1.214:502): Connection unexpectedly closed 1.0967254638671875e-05 seconds into read of 8 bytes without response from unit before it closed connection.

The adapter seems to receive the "init" frames and it looks like it can talk to the inverter but closes the connection prematurely:

Debug log

Logger: custom_components.foxess_modbus.modbus_controller
Source: custom_components/foxess_modbus/modbus_client.py:160 
Integration: FoxESS - Modbus ([documentation](https://github.com/nathanmarlor/foxess_modbus), [issues](https://github.com/nathanmarlor/foxess_modbus/issues)) 
First occurred: 12:05:17 (35 occurrences) 
Last logged: 13:33:27

Autodetect: failed to connect to (tcp://10.112.1.214:502)
Autodetect: failed to connect to (tcp://10.222.0.213:502)
Traceback (most recent call last):
  File "/usr/local/lib/python3.10/site-packages/pymodbus/client/tcp.py", line 309, in recv
    ready = select.select([self.socket], [], [], end - time_)
ValueError: filedescriptor out of range in select()

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/config/custom_components/foxess_modbus/modbus_controller.py", line 332, in autodetect
    await client.read_registers(
  File "/config/custom_components/foxess_modbus/modbus_client.py", line 73, in read_registers
    response = await self._async_pymodbus_call(
  File "/config/custom_components/foxess_modbus/modbus_client.py", line 160, in _async_pymodbus_call
    result = await self._hass.async_add_executor_job(call, *args)
  File "/usr/local/lib/python3.10/concurrent/futures/thread.py", line 58, in run
    result = self.fn(*self.args, **self.kwargs)
  File "/usr/local/lib/python3.10/site-packages/pymodbus/client/mixin.py", line 100, in read_holding_registers
    return self.execute(
  File "/usr/local/lib/python3.10/site-packages/pymodbus/client/base.py", line 192, in execute
    return self.transaction.execute(request)
  File "/usr/local/lib/python3.10/site-packages/pymodbus/transaction.py", line 159, in execute
    response, last_exception = self._transact(
  File "/usr/local/lib/python3.10/site-packages/pymodbus/transaction.py", line 306, in _transact
    result = self._recv(response_length, full)
  File "/usr/local/lib/python3.10/site-packages/pymodbus/transaction.py", line 341, in _recv
    read_min = self.client.framer.recvPacket(min_size)
  File "/usr/local/lib/python3.10/site-packages/pymodbus/framer/__init__.py", line 53, in recvPacket
    return self.client.recv(size)  # pylint: disable=no-member
  File "/usr/local/lib/python3.10/site-packages/pymodbus/client/tcp.py", line 311, in recv
    return self._handle_abrupt_socket_close(size, data, time.time() - time_)
  File "/usr/local/lib/python3.10/site-packages/pymodbus/client/tcp.py", line 363, in _handle_abrupt_socket_close
    raise ConnectionException(msg)
pymodbus.exceptions.ConnectionException: Modbus Error: [Connection] ModbusTcpClient(10.112.1.214:502): Connection unexpectedly closed 3.62396240234375e-05 seconds into read of 8 bytes without response from unit before it closed connection

[canton7]

First off, have you followed the configuration in https://github.com/nathanmarlor/foxess_modbus/wiki/Elfin-EW11?

[canton7]

What IP is what device? 10.112.1.214.502 is the EW11 presumably, and the unifi will be the router.

I'll look at the pcap at some point when I'm at home -- can't have wireshark on a mobile.

It looks like the router sends the modbus request (the packet with length 12), and then immediately sends a FIN.

However, I don't understand why everything is duplicated. Sure that's an artifact of how it was captured -- maybe we're capturing traffic as it crosses the bridge or something, those duplicate packets can't actually be being sent to the EW11.

[canton7]

Is HA running on unifi.lan?

[canton7]

Googling around, the problem might be that you're out of fds: pymodbus is calling select at that point, which fails if you try and give it an fd > 1024. It'd be interesting to see lsof | wc -l on the HA machine.

[canton7]

Yep, I'm pretty sure that's the problem. Looks like pymodbus has seen this error before, but mis-diagnosed it as the remote side terminating the connection early... Yay for python libraries 🙄

Looks like we can hack our way around it by moving that specific bit of functionality to use poll.

[canton7]

For now, try restarting -- that might clear out some FDs (as a temporary solution). Particularly if you're running HA on a unifi, I'd expect there to be a lot of active connections.

[canton7]

@mschae Can you give the latest pre-release v1.5.3b1 a spin?

I've hacked pymodbus to use poll() rather than select(). If I'm right in my diagnosis, that should fix the issue

[canton7]

@mschae Have you had a chance to try this out, by any chance?