Observatory
This project aims to make it easier to inspect events in redshift/postgres, imagine it like a lightweight kibana for redshift event data.
It provides an interactive ui that automatically queries and displays events.
It uses Google Charts to visualize events, and provides ways to automatically create simple graphs.
It also provides a way to generate documentation for events to markdown.
Installation & Running
Create a .env file with settings for your setup. Your Postgres/redshift connection details most importantly.
Then, you can run the whole stack with docker compose:
docker-compose up
You can also run each program by hand:
PG_HOST=XXX PG_PWD=YYY ... DEBUG=true python app/main.py
Web
A tornado web server and JQuery UI that serves up and graphs the event data. Depending on how you run it You can access it at http://127.0.0.1:6080/index.html or http://your-local-docker-host:your-port/index.html
The UI supports viewing a sampling of events, filtered by days back in time. It will pull all events down and build charts from them, so it won't scale well for large number of events.
You can use the buttons to build simple charts, split by zero or one field.
You can also aggregate charts into simple dashboards at /dashboards.html
Gen
generator.py is a program to generate markdown files from your event data. By default, every hour it will
generate events.md and events_dev.md, the latter for any tables with _dev as their suffix. It's not
yet very configurable, but it's a simple enough script that you could easily adapt. PRs welcome.
These files can be viewed in the UI at YOUR_HOST/docs.html and YOUR_HOST/devdocs.html respectively.
Kubernetes
The stack is well set up to slide into an existing k8s stack. Use the kube.yaml and service.yaml to
setup to your own need, and then proxy to behind some authentication lest your event data be publicly available to all.
Credits
observatory icon by Bernar Novalyi from the Noun Project
marked.js by chjj - https://github.com/chjj/marked