nix-cache01 regularly hits high CPU during nix builds, causing flappy
alerts. Keep the 15m threshold for all other hosts.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The step-ca serving certificate is auto-renewed with a 24h lifetime,
so it always triggers the general < 86400s threshold. Exclude it and
add a dedicated step_ca_serving_cert_expiring alert at < 1h instead.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add homelab.monitoring NixOS options (enable, scrapeTargets) following
the same pattern as homelab.dns. Prometheus scrape configs are now
auto-generated from flake host configurations and external targets,
replacing hardcoded target lists.
Also cleans up alert rules: snake_case naming, fix zigbee2mqtt typo,
remove duplicate pushgateway alert, add for clauses to monitoring_rules,
remove hardcoded WireGuard public key, and add new alerts for
certificates, proxmox, caddy, smartctl temperature, filesystem
prediction, systemd state, file descriptors, and host reboots.
Fixes grafana scrape target port from 3100 to 3000.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
