Project Description

supportconfig provides a lot of files and data from the system, but it is often hard to spot the real issue in it. The idea of this project is to get machine-readable output for the supportconfig data and analyze them, ideally with an existing tool.

Then we would try to provide hints using the tool about what is wrong.

Goal for this Hackweek

• Getting a machine readable version of supportconfig
• First analysis and tweaking

Summary:

• Research about machine learning log anomaly detectors: few alternatives out there.
• Getting custom metrics for Salt and Uyuni via prometheus exporter from live server.
• Setting up Loki to process relevant Uyuni logs from live server.
• Allow data visualization with Grafana.
• Really easy-to-use CLI tool to run "health checks" and get feedback.

Details:

• Loki and Uyuni prometheus exporter runs on "containers".
• The containers run on the Uyuni server. "podman" is required on the server.
• CLI tool takes care of building and deploying the "container" image to the server, collect the metrics and provide output on the command line.
• Prometheus / Grafana can be configured to expose containers metrics.

Some next steps:

• Enhance and collect more Uyuni / Salt metrics.
• Use "supportconfig" as source for logs/metrics instead of live server.
• Improve CLI and performance.

Resources

"Uyuni Health Check" GitHub repository: https://github.com/meaksh/uyuni-health-check/