Topics created by devmicha

Hello @devmicha,
The infrastructure looks like this:

IOs -> (SPI) -> ModbusTCP server -> (ModbusTCP) -> Evok -> (REST, WebSockets,...) -> your code

The ModbusTCP server (unipitcp service) is polling the complete status of underlying HW via SPI. This happens very quickly, in less than milisecond interval. The Evok aggregates data from the ModbusTCP server and from the owfs (for 1-Wire devices). Your application can poll data from the Evok in multiple ways (REST, WebSockets,...)

As you can see, the Evok doesn't communicate with the HW directly. We discussed the possibility of custom software pulling the SPI in this thread: https://forum.unipi.technology/topic/711/reading-spidev

I fully understand your concerns about infinite polling data over GET requests. This is generally slow and causes high load on the PLC. The WebSocket offers you a way to subscribe to updates of the IO. You can poll the complete status of the PLC at any time and yet you can listen to updates and act accordingly. Check the documentation and example: https://evok-9.api-docs.io/1.08/ubmjnyexu77wxwo4q/websocket.

Martin

You can change the

access_log /var/log/evok.access.log;

line near the top of your /etc/evok-nginx.conf file to

access_log off;

This should take care of it. You can also pipe it to /dev/null or create some sort of a log rotation scheme using CRON instead.

Jessie has a different netconf stack from Stretch, so the AP will not work there - and there is a rather large all-capitals warning during the installation to not use anything else than clean Raspbian Stretch.

Regardless it should work now, we'll have a quick look at what could go wrong.

@devmicha Its only possible to control one RO/DO with one DI with the same number using the DirectSwitch function. If you want to do 1Di to more Do/Ro, than you have to make some program which will take care of it. Or also just do it in the cables :)