Commit Graph
2 Commits
Author SHA1 Message Date
k.renner 3ec5802e62 monitoring: pin Loki resources to prevent OOM-kills
ci/woodpecker/cron/security-scan Pipeline was successful
ci/woodpecker/cron/e2e Pipeline failed
Chart ships no defaults for singleBinary.resources, which let the kernel
OOM-kill loki-0 on 2026-05-14 (exit 137 during initial install). Pin
requests at 100m / 512Mi and limit memory to 2Gi (~4x steady-state RSS
of 540Mi at current 37 lines/sec ingest). CPU stays unburstable.
2026-05-20 14:01:19 +02:00
k.rennerandClaude Opus 4.6 cfcd5048a6 feat: Deploy monitoring stack (Prometheus, Grafana, Loki, Alloy)
- kube-prometheus-stack: Prometheus (50Gi, 90d retention), Grafana
  with ingress at dash.renner.uno, Alertmanager, Node Exporter,
  kube-state-metrics. Scrapes existing Proxmox infrastructure.
- Loki: SingleBinary mode, 80Gi storage, 90d retention
- Alloy: DaemonSet collecting K8s pod logs to Loki
- Grafana datasources: Prometheus (default) + Loki
- Disabled etcd/scheduler/controller-manager scraping (Talos)
- Monitoring namespace set to privileged PodSecurity

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-05 17:27:01 +02:00