Should ReadHoldingRegistersReponse contain the address of the resister that was asked for?

[chrisbeardy]

I am using Read Holding Register and pass the address 550 for example. The repsonse is of type ReadHoldingRegistersResponse which contains the correct dev_id and accurate value of the registers and status=1. However the address is 0.

I am using one async call (asyncio.gather) to read multiple slaves with multiple different registers on each slave, so am looking for a way to easily relate the reply with what device and register. I was wondering if the response should contain the address or if I need to do something so it will? So far I have a workaround that I just read the same registers from all devices so can iterate through them using an index but would be nice to optimise this.

[janiversen]

That is correct, the address is not used in clients but only in server.

In a client you do not need to know it, because it is response to your request, so you have the address at hand.

[janiversen]

Asyncio gather, will not help you, pymodbus treats the call one by one, so first call is served and the rest waiting on an asyncio.lock

This correspond to the modbus protocol which is a request/response type.

[chrisbeardy]

Thanks for this, I did not realise that the modbus protocol meant that the Modbus server could not handle multiple requests at once. After some quick testing, I actually determined that in my case when using Modbus TCP and whatever devices I have, the sync client is actually quicker, and using async gather method actually the slowest while being the most cumbersome to code, therefore I will just go to using the sync client which will solve my problem of knowing the address much easier when decoding the return messages. Thanks fr your help.

[janiversen]

the async client WITHOUT gather is a factor faster than the sync client, but your choice.

[chrisbeardy]

It wasn't in my case from the testing I did in my environment. This may have been linked to the Modbus device I was reading data from though. I copied the example of async vs sync and changed the device slave etc to my device and ran the test. And there about the same speed.

[janiversen]

I was of course only talking about pymodbus.

If you run the performance test example, which uses a virtual device with no latency, you will see the difference. However it is true that many devices are quite slow, so the difference in pymodbus do not matter.

[chrisbeardy]

thanks for your time to reply to my questions, it been really helpful. This was the first time I've used pymodbus, so its good to have your insight. Thanks for your efforts!