Log causality graph: force-directed incident visualization (nodes=services, edges=correlated events) #31

New issue

Open

opened 2026-05-24 22:02:57 -07:00 by pyr0ball · 0 comments

pyr0ball commented

2026-05-24 22:02:57 -07:00

Owner

The current incidents view is a flat list with a linear timeline. A force-directed graph would make causality and blast radius immediately visible.

Proposed visualization:

Nodes: services/hosts (sized by event volume in window)
Edges: correlated log events between services (weighted by temporal proximity + shared pattern tags)
Color: severity (green → yellow → red) per node
Click: node expands to show contributing log entries
Time scrubber: replay incident propagation across the graph

Stack options: D3.js force simulation (fits existing Vue 3 SPA), or Cytoscape.js for richer graph controls.

This is a significant UI addition — suitable for a dedicated "Graph" tab on the incidents view.

Reference: https://github.com/Lum1104/Understand-Anything (force-directed knowledge graph for causality)

The current incidents view is a flat list with a linear timeline. A force-directed graph would make causality and blast radius immediately visible. **Proposed visualization:** - **Nodes:** services/hosts (sized by event volume in window) - **Edges:** correlated log events between services (weighted by temporal proximity + shared pattern tags) - **Color:** severity (green → yellow → red) per node - **Click:** node expands to show contributing log entries - **Time scrubber:** replay incident propagation across the graph **Stack options:** D3.js force simulation (fits existing Vue 3 SPA), or Cytoscape.js for richer graph controls. This is a significant UI addition — suitable for a dedicated "Graph" tab on the incidents view. Reference: https://github.com/Lum1104/Understand-Anything (force-directed knowledge graph for causality)