This repo provides my current work on a metric exporter for freqtrade.
The intended usage for this is to monitor multiple different running freqtrade instances, running different strategies.
You will be able to compare these against each other.
This whole repository is best used and intended to be used over the docker-compose.yml.
If you want to integrate the dashboard with your already present freqtrade instances, you have to add them to the same network the collector and prometheus resides in.
I always call this network the metrics network.
You have to create this shared network first by running docker network create metrics.
After this, you can run the example by running docker-compose up -d.
Inspecting the log output of the docker composition should indicate that everything is up and running.
Ideally, we would set the clients over docker-compose vars, but currently they are all defined in the collector/main.py file.
There, you'll find the BOTS dict, which holds a name used for identifying the bot in grafana and the hostname.
The hostname should be the container name of your freqtrade instance.
Your username/password HAVE TO be the defaults for every instance, which is freqtrader and SuperSecurePassword.
If you want to change this, you have to change the base64 encoded Authorization header, which currently representes the default username and password.
The freqtrade configuration should set listen_ip_adress to 0.0.0.0. The rest can be left as defaults.
After starting the instance, you should be able to visit grafana on localhost:8080. Setup your login credentials, then head over to the settings -> data sources. There, setup a new Prometheus data source. The URL is http://localhost:9090, the rest can remain as default values. Once done, you can head over to the dashboard import (http://localhost:8080/dashboard/import) and upload the here included FreqtradeDashboard.json. Now you'll see the (more or less) beautiful dashboard!
Currently, we make heavy use of the api over direkt requests. This isnt too pretty, as there already is a API client provided by freqtrade. In the future it would be nicer to switch to that API client.
The current application structure can be visualized like this:
