Question: Receive/send data over REST API about chirpstack-application-server HOT 6 CLOSED

I'll release the HTTP integration this week :-)

from chirpstack-application-server.

@crdn: why don't you setup a MQTT handler that would push radio stats into an influxdb database and then show graphs using grafana. I am doing this and this is really easy to setup.

In case you were talking about seeing the payload, that's application/vendor specific, so it doesn't make sense to handle this in the app-server stage.

from chirpstack-application-server.

Hi pmiam999

Thank you very much for your hint. As you mentioned, the bigger challenge is the conversion and interpretation of the vendor specific payload. So I set up an ELK Stack and I'm importing the data through MQTT into ElasticSearch. Logstash then helps me to convert the data and Kibana does the visualisation.

I worked with this (as far as I know unsupported) Logstash input https://github.com/widarlein/logstash-input-mqtt.

from chirpstack-application-server.

You should have a look at influxdb/grafana, this is a powerful and easy solution to store timeseries data. The decoding part can be a simple perl script listening to a given MQTT topics en pushing to influx (requires few lines of code).

from chirpstack-application-server.

Please follow #102, this will allow receiving data by setting up HTTP callback urls. The LoRa App Server format will be supported by MyDevices / Cayenne soon so that you can directly display your data in dashboard.

from chirpstack-application-server.

Circling back late to this, however there is now an (MQTT "Beat")[https://github.com/nathan-K-/mqttbeat] available for Elasticsearch that will pull data from an MQTT queue into Elasticsearch.

I'm using it successfully with The Things Network, I see no reason why it would fail to work with LoraServer.io.

InfluxDB is excellent (we're InfluxData partners we like it so much!), however if you want to scale out Influx properly (i.e. using more nodes rather than just throwing RAM at the problem!), then it becomes very expensive very quickly. Elasticsearch on the other hand scales out as well as up, and provides data resiliency "out of the box" as a result. (Full disclosure, we're also Elastic.co partners!)

If we only need metrics as opposed to log/event/metric correlation, then we'll use Influx. If we know that we'll need to scale the back-end storage and take advantage of multi-tier data retrieval speeds etc, then we'll use Elastic.

from chirpstack-application-server.