Your disks degrade for weeks before they fail. Prefail catches it — and alerts you in Telegram with what's wrong and what to do. SATA, NVMe, MegaRAID, Adaptec.
Try the DemoMost infrastructure monitoring is broad and shallow — 500 metrics per host, threshold alerts, email digests. Prefail does one thing: S.M.A.R.T. health, deep.
| Feature | smartd + email | Zabbix / PRTG | Datadog | Prefail |
|---|---|---|---|---|
| Scoring | Threshold only | Threshold only | Threshold only | 25+ attribute weighted score |
| Trend velocity | No | Manual setup | No | Auto: rising / stable / degrading |
| Hardware RAID behind controller | No | Limited | No | MegaRAID, Adaptec, mdadm |
| Time to deploy | 30 min/server | 2-4 hours | 15 min + $15/host/mo | 60 seconds |
| Weekly fleet digest | No | No | Custom build | Auto PDF every Monday |
| Cost at 50 servers | Free (setup cost) | Free (setup cost) | $750/mo | $39/mo |
A specialist tool replaces one part of your monitoring stack. It does not compete with Datadog — it covers what Datadog does not.
Open @prefail_bot in Telegram, tap Get Started, name your server. You get a unique token — paste it into the install command.
A single shell script. No Python, no Docker, no daemon to manage.
wget -O diskwatch.sh https://prefail.io/diskwatch.sh bash diskwatch.sh --token YOUR_TOKEN # ~60 seconds. Systemd service. 250 lines of bash.
Agent reports in 5 minutes. Alerts, weekly digests, and the full Mini App dashboard — all automatic.
Everything you need to manage disk health across your fleet — without leaving Telegram. Tap any server, see every disk, act on every alert.
Don't use Telegram? Email alerts available on all plans (Free: 1 recipient, Pro: 3, Business: 5, Scale/Enterprise: unlimited).
Every Monday, a PDF lands in your Telegram: which disks degraded, what changed, what to replace this week. Forward it to your team or print it for the rack.
Download sample reportEvery disk gets a 0–100 risk score. Not just threshold checks — SENTINEL weighs 25+ SMART attributes by statistical failure correlation, detects velocity of change (a counter growing 10×/day matters more than its absolute value), and classifies trends as rising, stable, or new. The same score gets different urgency based on whether the situation is accelerating or static.
Score history charts with 7-day deltas. Stable or rising classification per signal. See degradation velocity before it becomes critical.
Telegram push with email fallback. Each alert: severity, human-readable explainer, actionable recommendation. Ack, mute, or batch dismiss.
Toggle raw S.M.A.R.T. counters inline with disk rows. Reallocated sectors, current pending, CRC errors — all the numbers, always available.
MegaRAID 9260/9361/9460, Adaptec SmartRAID, Linux mdadm. Individual disk health behind the controller, BBU status, array state.
Full dark theme with pure-black AMOLED variant. Auto-detects Telegram scheme, manual toggle. CloudStorage-persisted preferences.
The agent is ~250 lines of bash. It runs smartctl, formats the output, and sends it via curl. Read every line before you install.
Collects only S.M.A.R.T. data. No file access, no network scanning, no process listing. Runs every 5 minutes, sleeps between.
TLS-encrypted channel with per-server API token bound to machine_id. Each server authenticates independently. No shared secrets between servers.
Data stored on a dedicated server in Europe. No third-party analytics, no tracking pixels, no data resale. GDPR-ready data export and deletion.
S.M.A.R.T. detects gradual degradation — not sudden failures from power surges, controller burnout, or electrical damage. Prefail is an early warning system, not a substitute for backups. Always maintain tested, off-site backups of critical data.
Every plan includes S.M.A.R.T. scoring, Telegram alerts, Mini App, and RAID support. Higher tiers unlock faster polling, longer history, and more email recipients.
Annual billing: ~17% off (2 months free). No credit card for Free.
Maintenance windows, REST API, webhooks, and team access are coming in Q2–Q3 2026 for Business tier and above.
At 200 servers with 1,000+ disks, expect 20–50 failures per year. Each one is 2 AM, a panicked Slack thread, and a scramble for replacement hardware. Or: a Telegram notification three weeks in advance. Prefail catches the ~80% of failures that are preceded by S.M.A.R.T. degradation. For the rest — power surges, controller failures, physical damage — maintain tested backups.
Your first 2 servers are free, forever. Start monitoring in 5 minutes.
Try the Demo