TELEICU/Middleware components
From OHC Network Wiki
More languages
More actions
The containerized middleware workloads that run on every K3s agent node at spoke and hub sites.
Core middleware
| Component | Role | Notes |
|---|---|---|
| Cloudflare Tunnel (cloudflared) | Secure outbound tunnel to Cloudflare edge | Credentials managed via sops; only outbound connection needed |
| TELEICU streaming service | Real-time patient data relay between spoke and hub | Core clinical middleware; latency-sensitive |
| Local image registry (zot / harbor) | Caches container images locally for offline resilience | Sites can operate from cache during internet disruption |
| Node monitoring agent | Classifies downtime cause (power vs network vs other) | Feeds into the incident notification chain |
| Vector / Alloy | Log shipping to centralized Loki/VictoriaMetrics | Buffered locally when offline, ships on reconnect |
| VictoriaMetrics agent (vmagent) | Metrics scrape and remote-write to central VM | Buffers on disk during disconnection |
Control plane services (cloud VMs only)
| Component | Role |
|---|---|
| K3s server (etcd) | Control plane with embedded HA etcd |
| VictoriaMetrics (single-node or cluster) | Time-series database for all 200+ node metrics |
| Grafana | Dashboards and alerting |
| Git mirror / webhook receiver | Trigger for configuration updates (comin / ansible-pull) |
| Incident manager | Routes alerts to hub admins based on downtime classification |
Standard container images
All middleware images are pinned to specific digests and served from the in-cluster registry or a private image repository. No site pulls from public registries at runtime — images are pre-cached for offline operation.