Networking Extension
The networking extension surfaces Cilium network policies and Hubble flow data inside ArgoCD. Developers can see which policies apply to their applications and observe actual network traffic.
Registration
The extension registers as:
- App view on any ArgoCD Application (detail page)
- Resource tab on core
Podresources
What it shows
Application view
When viewing an Application, the Networking view shows:
- A flow map: callers on the left, the app’s workloads in the middle, callees on the right. Pods are grouped by workload, external peers by DNS name or IP, and Cilium’s reserved identities (
kube-apiserver,host, …) get their own nodes. Calls between the app’s own workloads push the callee one column to the right, so a chain likefrontend → cart → checkoutreads left to right. Green edges carried only forwarded flows; red edges had at least one drop (dashed when nothing got through). Clicking a node or edge opens the flow table filtered to that peer. - CiliumNetworkPolicies applied to the application’s namespace
- CiliumClusterwideNetworkPolicies affecting the namespace (policies pinned to other namespaces stay hidden, so tenants don’t see each other’s rules)
- Cross-namespace traffic flows (via Hubble)
- Policy rule details (ingress/egress, endpoints, ports)
- For each flow, the policies Cilium reports as having allowed or denied it (expand a row in the flow table)
The map is built from the same flows as the table (up to 500 per refresh), so on very busy apps the counts are a sample of the window.
Pod resource tab
Per-pod network visibility:
- Active Hubble flows to/from the pod (the backend filters by pod before applying the limit, so a quiet pod in a busy namespace still shows its flows)
- Allowed and denied connections
- Source/destination details
Demo resources
The demo app includes:
- 2 application-level CiliumNetworkPolicies (in-namespace)
- 2 platform-level CiliumClusterwideNetworkPolicies
- 2 platform-level CiliumNetworkPolicies (cross-namespace)
- Cross-namespace traffic patterns for Hubble flow visibility
Backend
The networking backend queries Cilium/Hubble:
| Endpoint | Description |
|---|---|
GET /api/v1/policies | List network policies for an application |
GET /api/v1/flows | Query Hubble flows for a namespace or pod |
Configuration
| Environment Variable | Required | Default | Description |
|---|---|---|---|
HUBBLE_RELAY_URL | No | (empty) | Hubble relay URL (e.g. hubble-relay.kube-system.svc:80) |
PORT | No | 8082 | HTTP server port |
LOG_LEVEL | No | info | Log level |
CLUSTER_NAME | No | ArgoCD name of the cluster this backend reads from. When set, apps whose spec.destination.name differs get a 403. See “Multi-cluster ArgoCD” on the ArgoCD Configuration page. | |
CLUSTER_SERVER | No | API server URL of that cluster, matched against spec.destination.server |
RBAC
p, role:developer, extensions, invoke, networking, allow