---
title: Monitoring & Observability
category: systems
tags: [monitoring, prometheus, grafana, loki, holmesgpt, alerting]
created: "2026-04-28"
modified: "2026-07-24"
---

# Monitoring & Observability

## Prometheus + Grafana
- **Prometheus**: K8s (kube-prometheus-stack)
- **Grafana**: K8s, Ingress grafana.familie-schoen.com (.205)
- **Alertmanager**: K8s, Routes → Telegram

## Loki Stack (Centralized Logging)
- **Loki**: K8s ns=logging, LB 10.0.30.207:3100
- **Fluent Bit v5.0.9**: auf 9 PVE Hosts + 20 LXC + 5 QEMU + 6 K8s Nodes
- **Inputs**: PVE tasks, Ceph audit, journald, auth, pveproxy, pvedaemon, cluster
- Siehe [[systems/loki-fluentbit]]

## HolmesGPT (SRE)
- **Namespace**: holmes
- **Pipeline**: Alertmanager → adapter (9095) → HolmesGPT → TG @schoen_sre_bot
- **Model**: noris-glm (via ai.noris.de)
- RKE2 ctrl-plane alerts disabled

## Alerting Rules
- Via IaC Git Repo deployt (Ansible für CT141, ArgoCD für K8s)
- RKE2 control-plane alerts disabled

## Related
- [[systems/loki-fluentbit]]
- [[systems/rke2-kubernetes]]