Automating runbooks with Jupyrest
summary
Jupyrest turns Jupyter notebooks into REST APIs (using papermill and JSON Schema inputs), executes them via incident webhooks, and returns executed HTML while saving results as RDF/TTL graphs. Low-severity notebooks run automatically, producing small graphs that a high-severity notebook concatenates and queries with SPARQL to do root-cause analysis. Example pipeline stitches four low-severity runs into one SLA-dip investigation, identifies node 212 as culprit while node 900 is transient, and cuts time-to-mitigate in production deployments.
AI-written summary · about
chapters (9)
- 0:00Runbook automation with Jupyrest
- 2:40Automate investigation steps
- 4:28Jupyter to REST APIs
- 9:01Scalable runbook investigations
- 12:29Service domain modeling
- 14:16Low-severity incidents for high-severity investigations
- 19:08High-severity incident runbooks
- 21:17Root cause analysis with graph queries
- 24:05Key takeaways
related talks
31mNotebooks as Serverless Functions
34mData Products, From 0 to data science pipeline
28mFrom Jupyter Notebook to Eruption: Getting Geohazard Research into Production
01h 18mProductionizing Jupyter Notebooks
23mSecurity Automation Simplified
24mReliability Runs Efficiently at the Junction of Automation and Agile
39mNotebook To Production
17mR & Python playing nice, in Production