dumbmonit

DumbMonit is an actively developed, work-in-progress monitoring platform positioned as a simpler alternative to heavyweight tools such as Zabbix, Checkmk, or combined Prometheus/Grafana/Alertmanager setups, targeted at users managing around 10 machines, network switches, and self-hosted services. Its core design priorities are fast status readout, simplified device onboarding, and minimal alert noise. Key capabilities include: - Broad out-of-the-box data source support: SNMP v1/v2c/v3 for switches, routers, NAS, UPS, and printers with auto-applied MIB profiles detected via device sysObjectID; native integrations for Proxmox VE, Proxmox Backup Server, Proxmox Datacenter Manager, Proxmox Mail Gateway, OPNsense firewalls, Redfish-compatible server hardware (Supermicro, iDRAC, iLO, etc.), MDaemon email servers, and common self-hosted applications (Nextcloud, Immich, Paperless-ngx, Jellyfin, Plex, TrueNAS, Synology DSM). Cross-platform agents for Linux, macOS, FreeBSD, and Windows collect CPU, memory, disk, service, temperature, SMART health, and ZFS pool metrics, with one-line install scripts that register the agent with the host's service manager. Docker container monitoring is available via the agent, alongside service monitors for HTTP, TCP, DNS, ping, and TLS certificate checks, application-level session monitors for SMTP, PostgreSQL, MySQL/MariaDB, MQTT, and WebSocket, push-compatible heartbeat (dead man's switch) monitors, and CIDR network discovery for bulk device onboarding. - Low-noise alerting: Pre-built default rules cover common issues including unreachable devices, saturated CPU, near-full disks, UPS running on battery, stale backups, and expiring certificates. Features include dependency-based alert suppression (parent device outages suppress descendant alerts), host-based alert grouping and deduplication, configurable alert acknowledgement quiet periods, scheduled maintenance windows, notification policies with hysteresis, per-channel cooldowns, and quiet hours, seasonal baseline anomaly detection that learns network patterns without manual threshold tuning, predictive disk and datastore fill forecasts, and support for 22 notification channels including Discord, Slack, Microsoft Teams, Telegram, email, custom webhooks, PagerDuty, and Home Assistant. - Interface and operational features: Light and dark themes, full-screen wall mode for room displays, command palette navigation, mobile-optimized views for status checks and alert silencing, public branded status pages with 90-day uptime history, incident tracking, and email/RSS subscriptions. Access controls include admin/viewer roles, optional TOTP two-factor authentication, OIDC/SSO support with group-to-role mapping, and audit logging. The platform exposes scoped REST API tokens, a built-in MCP server for integration with AI assistants, Prometheus-compatible metrics endpoints for existing monitoring stacks, and encrypted configuration backup and restore functionality. Deployment runs as a single container, with embedded VictoriaMetrics for time series storage and SQLite for configuration and state, requiring no separate database services. Optional relay agents can be deployed at remote sites to run probes across networks without requiring inbound VPN access. As an alpha release, the platform is used daily on the author's homelab but expects rough edges, breaking changes, and no guaranteed upgrade path across database schema changes; some integrations have only been tested against simulated devices rather than real hardware.