26 opinionated alerts for what actually breaks: ECC errors, SMART failures, PSU redundancy loss, RAM pressure, disk saturation, network errors. One install. Five minutes.
IPMI sensors, ECC memory, PSU redundancy, CPU temperature
SMART health, NVMe wear, RAID status, disk latency, space usage
Interface errors, link speed, saturation, cable/SFP diagnostics
RAM pressure, swap usage, iowait, OOM kills, disk space
Self-hosted LLM analysis, plain-English findings, risk assessment
We spent a decade running servers across 60+ global locations. Every alert rule comes from a real failure we have seen. Not theoretical thresholds: real production breakdowns.
Crucible is MIT licensed. You can read every line of code that runs on your server. Forge adds the dashboard, history, and managed alerting on top.
Sign up, register a server, run one command. No YAML templates, no plugin configuration, no dashboard building. 26 alerts are active immediately.
Hosted on dedicated bare metal in Amsterdam. AI analysis runs on our own GPU, not third-party APIs. No tracking, no analytics cookies, no data sharing.
Everything in Free, plus:
Everything in Pro, plus:
Crucible runs on any Linux distribution with a standard kernel. Tested on Ubuntu, Debian, RHEL, Rocky, Alma, Arch, and Alpine.
Linux kernel 4.x or newer.
Crucible automatically detects and uses: smartmontools, ipmitool, mdadm, zfsutils. If a tool is not installed, the related checks are skipped.
One command. Detects your distribution, installs what it needs, starts monitoring. No configuration files to write.
