This repository contains rough proof of concept of a end-2-end events visualization across historical job runs (currently SNO 4.11 & SNO 4.11 upgrade). For now it only presents disruptions only.
Plotting e2e events trends could be useful in pinpointing when some problems started to occur.
- Data for specific jobs is fetched from GCP Bucket containing job run artifacts and stored to
workdir/orig/directory - The data is processed into per job JSON file that stores all runs with all disruptions.
- Processed data is exposed via http endpoints:
/jobs/- list of jobs/job/NAME- list of metrics available for that job/job/NAME/METRIC- array of job runs and data for specific disruption (total run duration, total disruption duration, disruption intervals)
/endpoint serves frontend file that fetches the data and presents a chart created with vega-lite specification
