initial: graph backend (client-go) + cytoscape.js topology frontend

Namespaces as compound parents, Deployment/StatefulSet/DaemonSet/CronJob
as workload nodes, Service/Ingress with edges derived from label-selector
matching and backend references, plus a physical-node layer via pod
placement. Click-through to live YAML manifest per node.
This commit is contained in:
2026-09-23 17:40:44 +03:00
commit 8f9c650960
7 changed files with 863 additions and 0 deletions
+20
View File
@@ -0,0 +1,20 @@
# cluster-map
Interactive topology view of the Project Chimera Deckhouse cluster --
namespaces group workloads (Deployment/StatefulSet/DaemonSet/CronJob),
connected to Services (by label-selector match) and Ingresses (by backend
reference), plus a physical-node layer showing what's actually scheduled
where. Click any node for its live manifest.
Go backend (client-go, builds the graph fresh from the k8s API on every
`/api/graph` request -- no caching, no watch loop, correct by construction)
+ a single-page Cytoscape.js frontend, no build step.
Phase 1 (current): plain in-cluster app, deployed like everything else in
this project (`03-kubernetes/apps/cluster-map/` in the `project-chimera`
docs repo, mirrored to `k8s-manifests` for Argo). No Hubble traffic overlay
yet, no packaging as an actual Deckhouse module (ModuleConfig/ModuleSource)
-- both are deliberately deferred until the base idea proves useful.
See STATUS.md / OpenProject (chimera-core) in `project-chimera` for the
full design rationale and decision history.