Feature Request

We are running multiple Rabbit MQ services which are monitorable via an API which in turn returns results via JSON.

Instead of creating 50 different REST sensors, would it be possible to define multiple channels on the same sensor?

An example of the JSON that comes out of the RabbitMQ service is

{"management_version":"2.7.1","statistics_level":"fine","message_stats":
{"publish":3371381,"publish_details":
{"rate":0,"interval":641379049,"last_event":1332232592280},"ack":2435,"ack_details":
{"rate":0.5845674201091193,"interval":5132000,"last_event":1332263428950},"deliver":3435,"deliver_details":
{"rate":0.5845674201091193,"interval":5132000,"last_event":1332263428950},"deliver_get":3435,"deliver_get_details":
{"rate":0.5845674201091193,"interval":5132000,"last_event":1332263428950},"redeliver":2731,"redeliver_details":
{"rate":0.0,"interval":5132000,"last_event":1332263428950}},"queue_totals":
{"messages":17014284,"messages_ready":17013284,"messages_unacknowledged":1000,"messages_details":
{"rate":-0.5845674201091193,"interval":338743284980,"last_event":1332263428950},"messages_ready_details":
{"rate":-0.5845674201091193,"interval":338743284980,"last_event":1332263428950},"messages_unacknowledged_details":
{"rate":0.0,"interval":338743284980,"last_event":1332263428950}},"node":"rabbit@NLRTD-RMQ01","statistics_db_node":"rabbit@NLRTD-RMQ01","listeners":
[{"node":"rabbit@NLRTD-RMQ01","protocol":"amqp","ip_address":"0.0.0.0","port":5672},{"node":"rabbit@NLRTD-RMQ01","protocol":"amqp","ip_address":"::","port":5672}],"contexts":
[{"node":"rabbit@NLRTD-RMQ01","description":"RabbitMQ Management","path":"/","port":55672}]}

There are multiple valuable parts of this JSON I would like to render in PRTG but now I have to write an intermediate (PHP) script.

Could this be added to PRTG?

Hi Megamuch,

we are currently looking into this issue and come back to you as soon as possible.

What i can say by now is that we think its an important feature, but we need to make sure that the configuration don't get too complicated.

Regards, Daniel

@Megamuch, can you please post your script? We are also preparing to monitor Rabbit, but 50+ sensors seems wasteful of probe and rabbit server resources.

Hi All,

I would love to see an example of settings / script for this sensor. I too am attempting to monitor RabbitMQ for various 'business' requirements (panic about queue depth, unacks due to flakey dev techniques on a web app), and have zero exposure to json / xpath etc.

When running queries against rabbitmq's api from curl, I almost get what I need. Translating this into the sensor has been a ton of frustration.

You can query every item of interest in a JSON document by specifying the node in an HTTP XML/REST Sensor. Add this sensor to PRTG and enter the URL to be requested for a JSON document.

Consider the example JSON code above taken from RabbitMQ. You can query items by providing, for example, the following entries in the XML Node (and optional property) field in the sensor settings:

Because the node “listeners” in the example code is an array (as like “contexts”, identifiable by square brackets), you need to address the elements of “listeners” by their number of appearance. “1” addresses the first element of the array, “2” the second one, etc. After the array element address, define the item to be queried of the specific array element (preceded by a slash sign).

To query the value of a node containing more than one element, or to query a value that is deeper in the JSON tree, provide the whole path (like for “last_event” in the table with example queries). Considering the example code, there are many “rate” values under different nodes. Providing the whole path ensures you to get the right value.

Important notice: You will have to add one sensor for each single value you want to query. Each sensor will send requests to the server. Thus, if you monitor a JSON file with, for example, 30 single sensors, then there will be 30 requests to the HTTP server with each scanning interval.