When choosing between Uptime Kuma and Beszel for homelab monitoring, the decision centers on what data you actually need. Uptime Kuma acts as an external endpoint checker that verifies if a service responds on a specific port or URL. Beszel operates as a lightweight host monitor that tracks internal server resources like CPU, memory, and disk utilization via installed agents. Because they solve different problems, experienced homelabbers run both side by side to catch application failures and resource exhaustion simultaneously.
Uptime Kuma: Endpoint Status Pages
Uptime Kuma determines if the target service is online. You configure it to hit an HTTP URL, a TCP port, or a DNS record. The server initiates a connection at a defined interval, logs the response time, and triggers an alert if the check fails. It handles public status pages and routes failure notifications to more than 90 supported platforms, including Discord, Telegram, and email.
Uptime Kuma operates externally, independent of the underlying operating system of the target application. You manage all configurations through the web interface. The default Docker container listens on port 3001.
Beyond standard HTTP status codes, Uptime Kuma executes keyword checks. This configuration downloads the HTML response and verifies that a specific text string exists. This catches scenarios where a reverse proxy returns a 200 OK status, but the application backend actually rendered a blank page or a PHP error trace.
For backup scripts and cron jobs, Uptime Kuma offers push monitors. Instead of the monitoring server initiating a connection, the tool generates a unique URL. Your remote script executes a simple curl command to this URL upon completion. If Uptime Kuma does not receive a request within the defined time window, it marks the job as failed.
Uptime Kuma Trade-offs
Uptime Kuma stores its configuration and historical response data in a local SQLite database mapped to /app/data inside the container. You must mount this path to a fast, local filesystem on the Docker host. The official documentation explicitly warns against storing this directory on an NFS or SMB network share. SQLite requires strict POSIX file locking to prevent database corruption. Network filesystems frequently handle these locks incorrectly, resulting in irrecoverable database failures.
Another common configuration issue involves Docker container monitoring. If you map /var/run/docker.sock into the Uptime Kuma container, the application reads the local Docker daemon to check container states. Passing socket access to a monitoring tool grants it root-level control over the host. A compromised Uptime Kuma instance can start, stop, or delete any container. A safer approach limits checks to HTTP and TCP requests across a Docker network bridge.
Beszel Monitoring: Host Resources
Beszel tracks a different metric: host performance and resource bottlenecks. Instead of pinging external ports, Beszel relies on a hub-and-spoke architecture. The central hub runs as a Docker container on port 8090. You then deploy a lightweight Beszel agent on every physical machine or virtual machine you want to track.
The agent extracts system metrics directly from the Linux kernel and the local Docker socket. It pushes CPU load, RAM usage, swap activity, disk I/O, and individual container resource limits back to the hub. The hub aggregates this data into a dashboard offering visual graphs.
Beszel requires strict key management. The hub does not poll random IP addresses on your network. When you add a new system in the Beszel web interface, the application generates a unique cryptographic key. You pass this key to the agent on the target machine using the KEY environment variable. The agent binds to port 45876 by default and waits for authorized requests.
This agent architecture ensures secure data polling. The agent uses the provided Ed25519 public key to verify incoming requests. If an unauthorized scanner hits port 45876 on your host, the agent drops the connection without returning any metrics. Bad actors cannot profile your server's hardware utilization.
Beszel Trade-offs
Beszel targets users who want quick visibility, not enterprise observability. It uses significantly less memory than a full Prometheus and Node Exporter stack. The hub container idles around 50MB of RAM, and the agent uses less than 20MB. This makes Beszel ideal for resource-constrained environments like older mini PCs or budget cloud virtual machines.
Beszel lacks an advanced query language. You cannot write custom queries to trigger alerts on compound conditions, such as sending a notification if disk space drops below 10 percent while CPU usage exceeds 80 percent for five consecutive minutes. The alerting engine supports basic, flat thresholds. If a host goes offline or memory exceeds a static percentage, the hub triggers a webhook.
Beszel stores its historical metrics in a local database managed by PocketBase, located in the /beszel_data directory. Like Uptime Kuma, you must mount this on a local filesystem to avoid lock contention. The hub retains data based on strict aggregation rules. It keeps high-resolution data for recent hours, then averages it into hourly or daily blocks for long-term storage to prevent infinite database growth.
Running Both in Your Homelab
Comparing Uptime Kuma against Beszel presents a false dichotomy. Endpoint checks do not warn you about a slow memory leak until the application crashes entirely. Host metrics do not indicate if an application threw a 500 Internal Server Error, because the host CPU and RAM might look perfectly normal during a software fault.
To achieve complete visibility, deploy them on the same host using a combined Docker Compose stack. Map the data directories to local paths and configure the restart policies to keep the monitoring stack online after a server reboot.
services: uptime-kuma: image: louislam/uptime-kuma:2 container_name: uptime-kuma restart: unless-stopped ports: - "3001:3001" volumes: - ./kuma-data:/app/data beszel-hub: image: henrygd/beszel:latest container_name: beszel-hub restart: unless-stopped ports: - "8090:8090" volumes: - ./beszel-data:/beszel_data beszel-agent: image: henrygd/beszel-agent:latest container_name: beszel-agent restart: unless-stopped network_mode: host volumes: - /var/run/docker.sock:/var/run/docker.sock:ro environment: PORT: 45876 KEY: "YOUR_PUBLIC_KEY_FROM_THE_HUB" In this deployment, Uptime Kuma monitors the HTTP responses of your proxy and the functional uptime of your services. Beszel monitors the disk space and Docker container limits of the host running those services. The Beszel agent binds directly to the host network interface to ensure accurate network throughput metrics. It requires read-only access to the Docker socket to report on individual container performance.
Because both tools use minimal resources, hosting them on a single node avoids impacting the performance of your primary applications. If your homelab spans multiple physical machines, keep Uptime Kuma and the Beszel hub on your primary node. Deploy just the beszel-agent service to the satellite machines.
When an application fails, you check Uptime Kuma to see exactly when the HTTP check dropped. You then check Beszel to see if the host ran out of memory or saturated its network link at the exact same timestamp.
Both tools provide web-based user interfaces. Exposing them directly to the internet on ports 3001 and 8090 leaves their login portals vulnerable to automated brute-force attacks. Place them behind a reverse proxy to handle SSL termination. Both applications rely heavily on real-time interface updates. If you configure a reverse proxy, you must explicitly enable WebSocket proxying, or the dashboards will fail to load live metrics.
What's Next
- Grafana + Prometheus Homelab Monitoring in 30 Minutes
- Docker Compose Homelab Stack: 20 Services, One File
- Fixing SQLite WAL Mode Reader Locks in Your Homelab
- A Minimal AWS Stack: VPS, DNS, Monitoring, Backups
FAQ
What is the main difference between Uptime Kuma and Beszel?
Uptime Kuma monitors external endpoints and network services to check if they are online. Beszel monitors internal host resources like CPU, memory, and disk usage via an installed agent.
Can Uptime Kuma monitor Docker container resource usage?
No. Uptime Kuma can check if a Docker container is running by mapping the Docker socket, but it does not track granular resource utilization like RAM or CPU usage. You need a tool like Beszel for resource metrics.
Do I need a reverse proxy for Beszel and Uptime Kuma?
Yes. Both tools provide HTTP web interfaces that should not be exposed directly to the public internet. A reverse proxy handles SSL termination and adds a layer of access control.
Originally published at linuxcore.dev.
