Collectd-web is a sleek, intuitive front-end for the powerful Collectd resource monitoring tool — visualize RRD metrics, monitor multiple hosts, and deploy in seconds.
Collectd is a high-performance system statistics daemon that has quietly powered production monitoring for sysadmins for more than fifteen years. It runs as a tiny native process on each host, samples hundreds of metrics — CPU, memory, disk, network, database internals, application counters — at sub-second granularity, and writes the results to round-robin databases on local disk. Because RRDs are fixed-size, storage requirements stay predictable for the life of the machine, which is exactly what you want when you're operating real infrastructure rather than building a demo.
What collectd never shipped, however, is a real dashboard. The bundled front-end was always treated by upstream as a testing tool — useful for confirming that data was being written, not for actually running operations from. Teams either wired up something heavy like Graphite or Grafana, hand-rolled their own RRD viewer, or just SSH'd into boxes and read text files. None of that is a great answer when you have a working, reliable agent that's already collecting everything you need.
Collectd-web is the dashboard the project always needed. It reads the RRDs collectd already produces, renders them with a modern UI, and stays out of the way the rest of the time. No new agent, no extra database, no SaaS account — just the missing front-end for an excellent backend.
Not every team needs a full observability platform. Collectd-web fits a specific shape of operator: someone who already trusts collectd, runs their own infrastructure, and wants graphs without the operational tax of a heavier stack.
If you operate a real fleet but don't want to stand up a Prometheus + Grafana + Alertmanager stack just to see CPU graphs, collectd-web is built for you. Drop the agent on each host, point the UI at the data directory, and you're done. There's no scrape config to maintain, no service-discovery layer to babysit, and no separate time-series database to back up.
Homelabs deserve nice graphs without enterprise overhead. Collectd-web runs comfortably on a Raspberry Pi, plays well with Proxmox VEs and TrueNAS jails, and gives you per-host visibility into the boxes you actually care about — without sending metrics off-box to a third-party SaaS.
When devices live behind flaky links or on constrained hardware, collecting metrics locally to RRD and viewing them remotely on demand is far more practical than streaming everything to a central time-series store. Collectd-web fits that pattern naturally — point it at a per-device data directory and you have a dashboard for the field.
Monitoring is a crowded space. Collectd-web doesn't try to replace every tool below — it's deliberately the smallest answer to one specific question: "I have collectd, how do I look at the data?" Here's where it sits relative to the obvious neighbours.
Grafana is a powerful, generic dashboarding layer that can read from dozens of data sources — and it expects you to bring those data sources, configure them, and operate them. For collectd users, that usually means standing up a Prometheus or Graphite pipeline alongside collectd just so Grafana has something to query. Collectd-web is purpose-built for collectd's RRD output and zero-config: install, point at the data directory, done. You give up Grafana's flexibility; you get back a stack that's an order of magnitude smaller to run.
Full comparisonCacti is the closest spiritual ancestor — an RRD-native web front-end with a long pedigree. The trade-offs are familiar: Cacti has accumulated a large feature surface (templates, user management, plugin architecture, polling subsystems) over its lifetime, and its UI shows its age. Collectd-web is intentionally smaller, ships a modern interface, and is Apache-licensed rather than GPL, which matters for some commercial deployments. If you need Cacti's built-in poller, stay with Cacti; if you want a thin, modern viewer for collectd-produced RRDs, collectd-web is the leaner choice.
Full comparisonNetdata is a beautifully presented, real-time monitoring agent that stores its own time-series data in its own format. That's a great fit for live troubleshooting on a single host, but it's a different architectural choice from collectd's RRD-on-disk model. Collectd-web stays RRD-native: it doesn't run an additional storage engine, doesn't need its own retention strategy, and integrates directly with the collectd you may already have deployed. Pick Netdata if you want an all-in-one agent; pick collectd-web if you've standardized on collectd and want a dashboard for it.
Full comparisonCollectd-web bridges the gap between Collectd's raw data and the dashboard your team actually wants to use.
Visualize system statistics with a clean, responsive design that works on any device.
Highlight the metrics that matter to you with a fully tweakable presentation layer.
Launch the built-in standalone Python server in seconds with minimal configuration.
Optimized for performance and usability across diverse systems and host counts.
Beautiful, accurate graphs straight from the round-robin database files Collectd produces.
Organize each host's data in its own subdirectory and browse them all from one place.
Pick your platform and copy the commands. Detailed setup lives in the docs.
# Download the latest .deb release
wget $(curl -s https://api.github.com/repos/httpdss/collectd-web/releases/latest \
| grep "browser_download_url.*\.deb" | cut -d '"' -f 4)
# Install
sudo dpkg -i collectd-web*.deb
# Start the service
sudo systemctl start collectd-webPer-host CPU usage and load averages over time.
Memory utilization and disk I/O at a glance.
Bandwidth, packets, and interface throughput.
Everything teams usually ask before deploying Collectd-web in production.
Release notes and deployment tips. No spam.
Collectd-web is open source and welcomes contributions. Star the repo, file an issue, or support development directly.