A real-time power outage & network monitoring tool built for COX's Bazar Cable Landing Station (COXCLS), operated by BSCPLC — watching grid power, generator, and ATS behaviour around the clock, passively and automatically.
⚠️ Passive monitoring only. SysMonitor has no direct connectivity to power infrastructure. It observes network reachability of existing monitoring points and infers power state — it never sends commands to the generator, ATS, or any power hardware.
SysMonitor is a lightweight, self-hosted power & network outage monitoring tool built specifically for COX's Bazar CLS (COXCLS), a facility that relies on grid power (PDB) with a generator + Automatic Transfer Switch (ATS) as backup.
It runs entirely on the local network. Instead of someone on shift needing to notice a power drop and log it manually, SysMonitor watches two always-on network devices and infers the facility's power state from their reachability — then keeps a permanent, timestamped record of everything, automatically.
- 🔍 Detects outages automatically — no manual logging, the system notices the moment power drops
- ⏱️ Tracks exact durations — PDB outage time and generator runtime, down to the second
- 📚 Permanent history — every outage cycle and event preserved for reporting and audits
- 📢 Instant alerts — Telegram, WhatsApp, and email notifications the moment something changes
SysMonitor is a Django web application paired with a lightweight background monitor process. The two run as separate, independent services and communicate only through the shared database — a restart of one never takes down the other.
- Background monitor pings two reference points — a "Holder" device (reflects grid/PDB power) and an "NVR" device (reflects generator/UPS power), pinged every few seconds
- State changes are cross-checked between the two — comparing both devices filters out false alarms like brief WiFi drops or a single device rebooting
- Confirmed changes are written to the database — outage start, generator start, power restore, and ATS switch-back are each timestamped
- The dashboard reflects it live — auto-refreshing status, live device health, and a running event log
- Notifications fire in parallel — the background monitor calls the notification dispatcher directly the moment a cycle-defining event happens
Three independent services, one shared database:
| Service | Job |
|---|---|
sysmonitor-ping |
Pings the Holder & NVR devices continuously, writes confirmed state changes to the database, fires notifications instantly |
sysmonitor-web |
The Django app — serves the dashboard, report, events, admin pages, and the public docs page |
sysmonitor-backup |
A timer that snapshots the database daily, so a bad write or disk issue never means losing history |
Holder Device ──┐
├──> Background Monitor ──> SQLite Database ──> Django Web App ──> Browser / Android App
NVR Device ─────┘ │ │
└──> Notification Gateways Backup Timer
(Telegram / WhatsApp / Email) (daily snapshot)
A full, healthy outage cycle: grid drops → generator picks up the load → grid comes back → ATS switches the load back to grid power.
| Cycle Type | Meaning |
|---|---|
| 🟢 Normal | Clean cycle — outage, generator ran, power restored, ATS switched back cleanly |
| 🔵 ATS-only | Just a brief ATS switch blip — not a real power outage |
| 🟡 Incomplete | An outage is currently in progress |
| 🟠 Alarm | Power was restored but generator/ATS didn't switch back as expected — flagged for review |
| 🔴 Critical | An extended total failure — both reference points unreachable for a long stretch |
| ⚪ Manual | Entered or corrected by hand via System Tools — for an outage the automatic monitor missed |
Why two devices instead of one? A single ping target is unreliable — a device rebooting or a WiFi hiccup can look identical to a real power outage. Requiring both the Holder and NVR to agree before confirming a state change filters out one-off blips automatically.
A maintenance mode can be switched on before planned network work, pausing outage detection and notifications so planned downtime is never logged as a false power outage.
- 📊 Live Dashboard — auto-refreshing status, device health, and daily outage summary
- 📋 Event & Activity Logs — full history of state changes plus an admin audit trail
- 🔧 Generator Shifting Entry — manual log of which generator is in auto mode and when
- 🛠️ System Tools — live internal state viewer, journal inspection, manual cycle correction
- 🧾 Reports — daily / historical outage & generator runtime reporting
- 👥 Role-Based Access — Admin, User, Viewer, and Guest roles, each with different visible pages
- 🙋 Self-Service Profiles — users manage contact details; sensitive fields need admin approval
- 💾 Automated Backups — database backed up daily, automatically
- 🔔 Multi-Channel Alerts — Telegram, WhatsApp, and email, configured per recipient
- 📱 Companion Android App — auto-detects local vs external network
- 🚧 Maintenance Mode — pause detection during planned network work
- ℹ️ Public Docs Page — readable by anyone, logged in or not, no secrets exposed
| Layer | Tech |
|---|---|
| Backend | Django (Python) |
| Database | SQLite, backed up automatically every day |
| Monitoring | Independent Python ping loop, decoupled from the web app |
| Frontend | Vanilla HTML / CSS / JS — no build step |
| Time handling | pytz — Asia/Dhaka, all timestamps normalised to local time |
| Notifications | Telegram Bot API, Meta WhatsApp Cloud API, Gmail SMTP |
| Process management | systemd services & timers |
| Mobile | Native Android wrapper (Java), WebView-based |
| Channel | Use case |
|---|---|
| 🤖 Telegram Bot | Instant push alerts to any number of subscribed chats |
| Business API-based alerts to configured phone numbers | |
| ✉️ Email | SMTP-based alerts for recipients who prefer email |
What triggers an alert:
- Outage start — sent once, the moment the Holder device goes down
- Critical — sent once if both reference devices are unreachable for an extended period
- Alarm — sent once per cycle if the generator or ATS doesn't behave as expected
- Cycle complete — sent once the cycle closes cleanly
Each fires at most once per cycle — the dispatcher tracks what's already been sent so recipients never get duplicate alerts for the same event.
To receive real-time Telegram alerts, message "Hi" to the SysMonitor notification bot: @SysMonitorCOXCLS_bot.
A lightweight Android wrapper app for quick access to the dashboard from a phone. It automatically detects whether you're on the facility's local network or an external connection and loads the right address.
Features: auto network detection · swipe-to-refresh · connection retry · in-app log viewer.
Installing requires enabling "Install from unknown sources," since it's a direct APK download rather than a Play Store listing. Dashboard login credentials are issued privately by the COXCLS NOC team.
| Role | Can see |
|---|---|
| Admin | Everything — Dashboard, Report, Events, Generator Shifting Entry, Devices, Users, Notifications, Activity Log, System Tools, plus profile-change approvals |
| User | Dashboard, Report, Events, Generator Shifting Entry |
| Viewer | Dashboard, Report, Events |
| Guest | Dashboard, Report |
When a user updates a sensitive profile field (email, mobile), the change is held as pending until an admin reviews and approves or rejects it, right inside the admin's Users tab.
Does SysMonitor control the generator or ATS? No. It's strictly passive — it only pings network devices to infer power state.
What's the difference between Viewer and Guest? Both see Dashboard and Report. Viewer additionally sees the Events log; neither sees Generator Shifting Entry or any admin page.
How are false alarms filtered out? By requiring both the Holder and NVR devices to agree on a state change before it's confirmed.
What happens during planned network maintenance? Turn on Maintenance Mode first — it pauses outage detection and notifications for the duration you set.
How often is the database backed up? Automatically, once a day, via a dedicated systemd timer.
How do I get an account? Accounts are created privately by the COXCLS NOC team, who also assign roles.
MD. Jikrul Sayeed Hossain Developer · KUET, EEE (2K15) · COXCLS NOC, BSCPLC
Designed, built, and maintains SysMonitor end-to-end — the Django backend, outage-detection logic, notification system, and companion Android app.
© 2026 COXCLS NOC, BSCPLC. All rights reserved.