Parse JSON response

[lhirdman]

Hi!
First time poster here, not that experienced with Zabbix. I'm currently working in a project to migrate from Shinken/Nagios to Zabbix 4.2.

I'm monitoring some applications with a health check page returning health status of application and also showing dependencies. I have tried to get both the health status and the message from the page to show up in the same item or trigger to no avail.
The item I created is an HTTP agent check and the format of the health check page looks like this:

{
    "dependencies": [
        {
            "isHealthy": true,
            "message": "App1 is connected and configured correctly",
            "name": "App1 name"
        },
        {
            "isHealthy": false,
            "message": "App2 is b0rked",
            "name": "App2"
        },
        {
            "isHealthy": true,
            "message": "App3 is connected and configured correctly",
            "name": "App3"
        },
        {
            "isHealthy": false,
            "message": "App4 doesn't like your code",
            "name": "App4 name"
        }
    ],
    "isHealthy": false,
    "message": "App2 and App4 failed",
    "name": "MainApp",
    "version": "1.1.4-prod"
}

What I would like to do is to at least be able to monitor the Health status of main app ($.isHealthy) and in the same time publish the content of $.message.
When I create an item that reads the output from the health check and add a preprocessor to it any subsequent preprocessor inherits the output of the previous preprocessor so I can't extract more than one key, value pair that way.
Is it possible to create macros based on different part of the json so I could use them when describing the problem in a trigger?
I have googled around but not found anything substantial only to use external scripts or userchecks which I will do if there's no way to extract multiple fields from the json into the item or a trigger.

[troffasky]

You want an HTTP agent item first to contain the entire blob of JSON.
Then for each parameter you want to extract, create a Dependent Item with your HTTP agent item as the Master. Give it a JSON Path preprocessing step, something like $.dependencies[0].isHealthy for App1 [or are they numbered from 1? can't remember].

If somebody knows an easy way to populate [or, "low level discover"] all the potential dependent items with preproc steps, I'd like to hear it.

Edit: just re-read your post. Seems like you actually already know how to do the JSON Path bit.

[Hamardaban]

The key phrase is: create as many dependent items, how many metrics you need to track.

[gert.derouck]

For the dependent items, you could use LLD with JSON Path preprocessing step and LLD macro's.

https://www.zabbix.com/documentation/current/manual/discovery/low_level_discovery

Low-level discovery now accepts optional user-defined LLD macros with a custom path specified in JSONPath syntax.

Example: I use this for discovering Jenkins jobs via its API.