Re do/enhances calculating expected response for various protocols

[dhoomakethu]

ModbusTCP seems to be having a problem reading everything in the response, refer #217, #200 , #202

Fixes for slwo reads ot the following protocols:

• TCP
• RTU
• Serial

[izak]

I seem to be having a problem related to this issue. If not helpful, I apologise. I'm trying to communicate with an SMA solar inverter using the Sunspec protocol. The response is supposed to be 0x53 0x75 0x6e 0x53 (literally SunS), but it seems the last byte is chopped off and then I get the same No Response error.

import logging
from pymodbus.client.sync import ModbusTcpClient as ModbusClient

def main(host, port, unit):
    logging.basicConfig()
    logging.getLogger('pymodbus').setLevel(logging.DEBUG)
    with ModbusClient(host, port=port, timeout=20) as client:
    print client.read_holding_registers(40000, 2, unit=unit)

if __name__ == "__main__":
    main('192.168.0.131', 502, 126)

And the output:

DEBUG:pymodbus.transaction:Running transaction 1
DEBUG:pymodbus.transaction:send: 0x0 0x1 0x0 0x0 0x0 0x6 0x7e 0x3 0x9c 0x40 0x0 0x2
DEBUG:pymodbus.transaction:recv: 0x0 0x1 0x0 0x0 0x0 0x7 0x7e 0x3 0x4 0x53 0x75 0x6e
DEBUG:pymodbus.transaction:0x0 0x1 0x0 0x0 0x0 0x7 0x7e 0x3 0x4 0x53 0x75 0x6e
DEBUG:pymodbus.transaction:getting transaction 1
Modbus Error: [Input/Output] No Response received from the remote unit

[dhoomakethu]

@izak, could you please give a try with the code on dev branch ? It is supposed to be having a fix with respect to modbus TCP.

[izak]

This happens with the code on the dev branch too. That was the first thing I tried, because I noticed some work had gone into this already.

The trouble seems to be that socket.recv receives a byte short, which is perfectly legit, that's how the unix syscall works. But because we're running a SocketFramer, it does something else I don't quite yet comprehend. I'm only about 2 hours into the code so I don't yet fully understand how it is designed.

[dhoomakethu]

Ok, if that is the case could you please try the read transaction twice and see if the missing byte is cleared before the second read transaction ? Please share the logs here.

[izak]

By literally just calling self._recv a second time in transaction.py and appending the result, it does indeed work.

DEBUG:pymodbus.transaction:Running transaction 1
DEBUG:pymodbus.transaction:send: 0x0 0x1 0x0 0x0 0x0 0x6 0x7e 0x3 0x9c 0x40 0x0 0x2
DEBUG:pymodbus.transaction:recv: 0x0 0x1 0x0 0x0 0x0 0x7 0x7e 0x3 0x4 0x53 0x75 0x6e 0x53
DEBUG:pymodbus.transaction:0x0 0x1 0x0 0x0 0x0 0x7 0x7e 0x3 0x4 0x53 0x75 0x6e 0x53

[dhoomakethu]

Just in case useful, I found an example for sunspec client here https://github.com/riptideio/pymodbus/blob/29d236edaccd48ddc35b5503298d157be2428cd1/examples/contrib/sunspec_client.py, the example is very old and might be buggy, give it a try.

[izak]

My use case was really to poke around a bit to see what I have to work with. I'm actually working on a project with code written in C, that is, I'm not actually going to use pymodbus in this project (but we are using it in this one, where it works fine). I since came up with a different way to poke around: Just sniff the traffic from the wire with wireshark. So I'm probably good to go.

[dhoomakethu]

Cool, I will keep this case in mind while fixing this issue, thanks for bringing this up.