Latest LLM Summary

## Daily summary
• **RouterOS Capacity**: CPU 2.0% (Δ 0.0%, 1.4x 7-day avg 1.43%), Memory 15.22% (Δ 0.0%, stable vs avg 15.12%). Uptime 9.76 days. No capacity pressure.
• **RouterOS Security/Network**: 20+ SYN probes logged to ports 80/443 from external IPs (05:47–06:04 UTC / 08:47–09:04 EET); no firewall drop spikes. WireGuard peer3 experienced transient handshake failures (08:51–08:54 EET) but recovered; all 3 peers online.
• **Docker Reliability**: 45/45 containers running, 0 unhealthy, 0 restarts. Total error lines 309 (Δ 0.0, below 7-day avg 364.71). `phase-backend` (137) and `personal-startpage` (57) lead error counts but remain within baseline.
• **Proxmox Capacity**: Nodes `proxmox` and `proxnode` memory at 80.7% and 80.8% respectively (Δ 0.0%, +4.3% above 7-day avg 76.47%). Exceeds 80% threshold; monitor for swap pressure or OOM events.
• **Proxmox Workloads**: 6 VMs and 2 LXC containers running. 6 VMs stopped (`lemonade-ubuntu`, `llama-vulkan-bookworm`, `cachyos`, `win11-gpu`, `llm-dual-gpu`, `win11-pro`). No unexpected state changes.
• **Proxmox Storage**: Max utilization 73.1% on `nvme1tb` (Δ 0.0%, matches 7-day avg). All pools under 80% threshold. Disk health PASSED across all nodes.
• **PBS Backups**: No backup telemetry received; last successful backup and failure status unknown.

## Actions
- proxmox: check memory pressure & swap: `ssh proxmox 'free -h && cat /proc/meminfo | grep -i swap'`
- proxmox: verify PBS backup configuration: `ssh pbs-server 'pbs status'`
- routeros: monitor WG peer3 handshakes: `ssh routeros '/interface/wireguard/print'`

## FYI (normal but noteworthy)
• RouterOS interface `wireguard1` shows 56,203 cumulative TX errors; stable vs yesterday, likely protocol-level noise or MTU fragmentation.
• Docker host resource metrics (CPU/RAM/Disk) not reported in payload; container-level health remains nominal.