A HTTP API based on Flask for serving pandas data frames to Grafana, generated by NumPy. The Grafana Simple JSON Datasource is used to interface Grafana with the HTTP API.
This way, a pure Python application can be used to directly supply data to Grafana, both easily and powerfully.
The framework supports feeding both timeseries data as well as annotations
through corresponding /query and /annotations endpoints and also
provides /search and /panels endpoints.
Image: Sinewave data and midnights annotations, both generated using NumPy, see Sinewave/Midnights example.
Grafana pandas datasource setup outlines how to install all software prerequisites needed to run this project. Please read this section carefully.
pip install grafana-pandas-datasource
Test drive:
# Run Grafana pandas datasource demo.
python examples/sinewave-midnights/demo.py
# Submit a timeseries data request.
echo '{"targets": [{"target": "sine_wave:24"}], "range": {"from": "2022-02-22T15", "to": "2022-02-22T20"}}' | http http://127.0.0.1:3003/query
# Submit an annotation data request.
echo '{"annotation": {"query": "midnights:xx"}, "range": {"from": "2022-02-20", "to": "2022-02-22"}}' | http http://127.0.0.1:3003/annotations
When the environment has been properly configured, both requests above will yield appropriate responses.
Then, configure the Grafana entities. You will need a datasource object and a dashboard object.
There are different demo programs accompanied with Grafana datasource and dashboard definition files.
After confirming the sandbox environment has been installed successfully, please head over to the Sinewave/Midnights example page in order to learn how to provision Grafana with corresponding resources.
In order to conceive your own pandas-based data source, please use the Sinewave/Midnights demo.py as a blueprint. If you think it would be a valuable contribution to the community, we will be happy to add it to the repository.
Kudos to Linar, who conceived the initial version of this software the other day at https://gist.github.com/linar-jether/95ff412f9d19fdf5e51293eb0c09b850.
Oz Tiram conceived a similar piece of software with Python. He uses the Bottle web framework.