In certain special cases it might be desired to avoid placing any middleware like an MQTT or a home automation server just to transfer small bits of data from one node to another. Direct data polling is possible using HTTP, but beware that the involved components are resource hungry and may be less stable on long term. The webserver embedded in the node is not designed to constantly serve a large amount of requests.
Setting up a webserver using the :doc:`/components/web_server` on the primary node will make available the required sensor data through a :ref:`api-rest` interface.
On the client nodes we need an :doc:`/components/http_request` with an ``id`` set, and a :doc:`/components/sensor/template` to make it accessible locally.
To automate the request for data, we use an :ref:`interval` requesting the URL pointing to the sensor id for which the state is needed. See :ref:`api-rest` on how to build up the URL for your sensors.
In the example below we request the value of a sensor from the server node, and after parsing the resulted JSON string we publish it to the local template sensor:
For security reasons, it's always recommended to protect the web interface of the nodes with authentication, even if you're using them on your local network.
Server part
***********
Add authentication to the ``web_server`` component on the primary node:
Add an ``Authorization`` header to your ``http_request.get`` action. The simplest way to determine a working authorization header is to visit the password-protected REST URL of the primary node using a browser while watching the network traffic in the browser's developer tools. If you look at the headers of the request sent by the browser, you'll find the ``Authorization`` header it sends to the node, and you can copy it for your own replay:
