2GT-RichandClaude Opus 5
d9fa120abc
Follows #137. The v0.24.1 fix made the count honest and stopped a thousand round trips per poll, but it assumed the unique index on (device_id, log_id) would reject lines already stored. For most devices it does. For a device whose log entries carry no `.id` it never could: the column was set to NULL, and PostgreSQL treats NULLs as *distinct* in a unique index, so every NULL conflicts with nothing. Those devices re-inserted their entire buffer on every poll, for ever, and the constraint that was supposed to stop it was structurally incapable of firing. A line with no id now gets a key derived from its own content -- timestamp, topics and message -- so the existing index does the work. The `#` prefix keeps it out of the space of real RouterOS ids (`*` plus hex), so the two cannot collide, and parseRosId already rejects it, which matters because the watermark is a max over real ids and must not be poisoned by a synthetic one. The trade, stated rather than buried: two genuinely separate lines sharing a timestamp, topic and message collapse into one. RouterOS timestamps are second-resolution, so that can happen. Losing a repeat of an identical message is a much smaller loss than growing the table by the whole buffer every minute without bound. This was found by reading a comment on #137 I had not read before closing it -- log output from a device with eleven lines total, reporting eleven new entries every minute for forty-five minutes. My own fleet cannot reproduce it: every event here carries a real `.id`, with no nulls and no duplicates, which is exactly why the hole was invisible from here. So this closes a defect demonstrable from the schema; whether it is the whole of that reporter's problem is still an open question and the issue is reopened to ask. Verified no regression where ids are healthy: no surrogate keys generated, no nulls, no duplicates, and counts that vary with real activity. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
26 KiB
MikroTik Manager
A self-hosted control plane for your entire MikroTik fleet.
Monitor, configure, and safely change routers, switches, and access points — from one web interface.
Contents
- What it is
- What makes it different
- Screenshots
- Get running in five minutes
- Features
- Documentation
- Contributing
- License
What it is
MikroTik Manager gives a whole fleet of RouterOS devices a single management surface: live monitoring, client tracking, VLAN and wireless configuration, firewall management, firmware rollouts, backups, and traffic analytics — without logging into WinBox on each box in turn.
It runs entirely on your own hardware via Docker Compose, talks to devices over the RouterOS API (and SSH where RouterOS requires it), and stores nothing outside your network.
What makes it different
Plenty of tools can show you a MikroTik fleet. The hard part is changing one safely.
RouterOS will happily lock you out of your own device
RouterOS applies every command immediately and independently. There is no transaction, no rollback, and no cross-object validation — so it will accept a change that severs the very path you are managing it over. MikroTik's own documented advice for enabling VLAN filtering is, literally, to have a serial console ready.
That is the problem this project takes seriously. Three layers address it, and they are the reason this platform exists rather than another dashboard.
🛡️ Change Guard — the device rescues itself
Before applying a risky change, the device saves a restore point and arms a scheduler to reapply it. The change goes in, then reachability is proven on a brand-new connection. Confirmed, and the scheduler is disarmed. Unreachable, and the device restores itself and comes back — unattended, with nobody driving to the site.
Verified end to end on a CRS running RouterOS 7.23.3 by deliberately deleting a switch's own management address:
T+0 restore point saved, revert scheduler armed, change applied
T+0s device drops off the network
T+40s manager gives up after 4 fresh connection attempts and reports back
T+120s device restores its own backup and reboots
T+3m device is back, change undone, scheduler and restore point gone
It runs entirely over the RouterOS API, so it works on devices with no SSH credentials configured, and it covers twelve change types — bridge VLAN filtering, port PVID and tagging, bridge VLAN add/update/ delete, IP address add/remove, route add/remove, bond create/delete, and management service toggles.
🔮 Lockout prediction — a warning that names the actual mechanism
Rather than blocklisting operations someone once thought were dangerous, the platform reads live device state, works out how the manager actually reaches the device, simulates your change against it, and reports any invariant that flips from satisfied to violated.
The ingress port comes from the bridge forwarding table and the manager's own address from the device's connection tracking, so the warning is specific enough to act on:
This change is predicted to cut management access to 2GT-NW-100G. Management arrives untagged on
sfp28-1(PVID 1) — the gateway's MAC is learned there — but VLAN 1 has no bridge VLAN entry listingbridge1as an untagged member.
A predicted lockout is refused outright. Overriding a critical verdict means typing the device name, and the change still runs under Change Guard.
🩺 Config Health — finds what RouterOS accepted but never applied
A standing, read-only audit for configurations RouterOS takes without complaint and then quietly
ignores. It encodes MikroTik's documented Layer 2 misconfigurations — an IP address or VLAN interface
on a bridge slave port, a bond slave that is also a bridge port, MTU above L2MTU, a PVID that
frame-type=admit-only-vlan-tagged renders inert, management surviving only on a dynamic VLAN entry,
and more.
Every finding explains what it does to your network, how to fix it, how long it has been there, and links the relevant MikroTik documentation.
And the rest of the toolbox
| Staged firmware rollouts | Canary waves, pre-upgrade backup, reboot verification, and halt-on-failure so a bad build never reaches the fleet |
| Built-in NetFlow collector | Traffic analytics with per-client attribution and application breakdown — no external collector to run |
| CAPsMAN aware | Centrally provisioned APs are recognised, grouped under their controller, and protected from local writes the controller would discard |
| Rogue AP detection | A foreign BSSID broadcasting one of your SSIDs is flagged as an evil twin |
| Anomaly insights | Client counts and CPU compared against each device's own same-hour 14-day baseline, not a fixed threshold |
| Multi-site | Group devices into sites and view each one's clients, events and topology separately, with an all-sites map. One install, many customer networks |
| Enterprise auth | OIDC/SSO, TOTP two-factor, role-based access, and scoped API tokens |
Screenshots
 Device list |
 Device overview |
 Switch ports & throughput |
 Hardware monitor |
 Network topology |
 Wireless radios |
 Client tracking |
 Client detail |
 Event log |
 Backups |
 Sign-in, with optional SSO and two-factor |
|
Get running in five minutes
You need Docker with Compose v2, and MikroTik devices running
RouterOS 6.x or 7.x reachable on their API port (8728, or 8729 for API-SSL).
Option A — pre-built images (recommended)
No source checkout, no build toolchain.
# 1. Grab the compose file and environment template
curl -O https://raw.githubusercontent.com/2GT-Media-Group-LLC/mikrotik-manager/main/docker-compose.ghcr.yml
curl -O https://raw.githubusercontent.com/2GT-Media-Group-LLC/mikrotik-manager/main/.env.example
mv .env.example .env
# 2. Set your domain in .env (secrets auto-generate if you leave them unset)
# CORS_ORIGIN=https://manager.example.com
# 3. Start
docker compose -f docker-compose.ghcr.yml up -d
Open https://localhost (or your server's hostname) and sign in with admin / admin.
Change that password immediately under Settings → Users.
Updating:
docker compose -f docker-compose.ghcr.yml pull && docker compose -f docker-compose.ghcr.yml up -d
Option B — build from source
git clone https://github.com/2GT-Media-Group-LLC/mikrotik-manager.git
cd mikrotik-manager
cp .env.example .env # edit CORS_ORIGIN for production
docker compose up -d
First run builds the frontend and backend, initialises PostgreSQL and InfluxDB, and generates a self-signed TLS certificate. Accept the browser warning, or upload a real certificate under Settings → TLS Certificate.
Updating:
git pull && docker compose up -d --build backend nginx— migrations run automatically on startup.
If a device won't connect
API access is usually enabled already. If not:
/ip service enable api # port 8728
/ip service enable api-ssl # port 8729, TLS
Devices on the SSL port are reached over TLS automatically — RouterOS's default self-signed certificates are accepted, so credentials are encrypted in transit with no manual trust step.
Features
Safety and change protection
The layers described above, plus the surrounding machinery.
- Change Guard — device-side auto-revert on twelve change types, over the RouterOS API, no SSH required
- Lockout prediction — live-state simulation that refuses a change predicted to sever management, with the hop chain that explains why
- Config Health — scheduled audit for RouterOS's documented Layer 2 misconfigurations, surfaced on the device's Security tab and in the dashboard
- Firewall lockout guard — refuses an unscoped
input-chain drop/reject that would lock you out, requiring explicit confirmation - CAPsMAN write protection — a provisioned radio is owned by the controller, so local SSID creates, edits and deletes are refused rather than silently discarded
- Capability probe —
POST /api/devices/:id/change-guard/probereports what safety mechanisms a device actually supports, tested rather than assumed, and leaves nothing behind - Guard history — every protected change recorded per device as
committed,reverted, orfailed
For the full behaviour — what the API returns when contact is lost, how binary and script modes differ, and how to choose a timeout — see Change Guard and Config Health.
Monitoring and operations
- Live dashboard — device counts, wireless clients, active alerts, fleet-wide 30-day availability, device-type distribution, firmware notifications, and a historical client graph (1h → 30d)
- Operations view — a second dashboard mode built for running the network:
- Things to handle — a server-side insights engine, severity-ordered: offline devices, pending firmware, CPU/memory pressure, missing backups, connectivity flapping, WiFi quality problems, and Config Health findings
- Quick actions that actually run — discovery, back up all online devices, sync config, or open an in-browser SSH terminal, each with inline progress
- Capacity & health — per-device CPU and memory meters sorted by pressure
- Security posture rollup — fleet hardening score and the lowest-scoring devices
- Anomaly insights — each device's last 30 minutes of client count and CPU compared against its own same-hour-of-day 14-day baseline, flagging ≥2.5σ deviations plus error-log bursts
- Availability tracking — per-device uptime %, outage count, and longest outage over 30 days
- Alerts — device up/down, CPU/memory thresholds, certificate expiry, firmware available, log errors and warnings, new device discovered, and config drift, each with cooldowns
- Delivery channels — Email, Slack, Discord, Telegram, and ntfy (self-hosted or
ntfy.sh, with severity mapped onto ntfy priority so outages break through do-not-disturb and recoveries stay quiet) - Maintenance windows — one-time or recurring (cron) windows per device that suppress alerts automatically
Switching, VLANs and wireless
- VLAN management — create, edit and delete VLANs; per-port membership with tagged/untagged control
- Per-port connected clients — selecting a port shows who is physically on it. Uplink and trunk ports are auto-detected (via an LLDP/MNDP neighbour, MACs spanning multiple VLANs, or a high MAC count) and show an explainer rather than every MAC reachable through them, with one-click disclosure of the full table
- Copy VLANs between switches — a three-step wizard with click-to-cycle port assignment, conflict detection, and a review summary before anything is applied
- Wireless management — SSID create/edit/enable/disable/delete, bulk SSID deployment across APs, WPA2/WPA3 security profiles, and support for both the RouterOS 7
wifipackage and the legacywlanpackage - RF Health — channel usage across 2.4/5/6 GHz with co-channel overlap highlighting, AP deployment density plotted against RSSI, TX-retry histograms, and a connectivity funnel (association → authentication → DHCP) derived from device logs
- Spectral and AP scans — scheduled or on demand, per radio
- Rogue & neighbour AP detection — stored scan results cross-referenced against your own SSIDs and radio MACs; a foreign BSSID broadcasting your SSID is flagged as an evil twin
- Device fingerprinting — every client classified into a category (server, computer, phone, TV, camera, printer, console, IoT…) from OUI and hostname, overridable per client and persistent across polls
- CAPsMAN — role detection (standalone / CAP / controller / controller-with-local-radios), provisioned interfaces labelled with their controller, and a controller panel grouping every AP with its radios, live channel, provisioned SSID and client counts. CAPs are matched to managed devices by radio MAC — hardware identity, so it cannot confuse two devices on two segments. Read-only for now; see CAPsMAN scope
- Guest WiFi — a guided wizard that builds a full captive portal in one pass: guest SSID on every radio, VLAN segregation, IP pool, DHCP, hotspot profile, bandwidth-limited user profile and optional NAT. Plus batch vouchers with printable 3-up sheets, a live guests-online table, and an inline walled garden
Firewall and network services
A Meraki/UniFi-grade firewall experience on the full RouterOS feature set.
- Security Center — fleet-wide hardening scores, per-device posture, and a "Common Findings" rollup aggregating identical issues across every device
- Friendly rule builder — action chips, address and address-list pickers, well-known port presets, connection-state chips, per-rule logging, and a plain-English preview of every rule
- Address lists as reusable objects — define
LAN,Trusted,Blocklistonce and reference them anywhere - Rule reordering — order is decisive in RouterOS; move rules with one click
- Hit counters — per-rule packet and byte counts surface which rules match and flag dead ones
- NAT wizards — guided Port Forward, Masquerade and 1:1 NAT flows, plus a Custom mode
- Security posture audit — per-device hardening checklist with a score and one-click remediation
- Bandwidth control — simple queues per IP, subnet or interface, with a one-click "limit this client"
- Active connections — live connection-tracking table with search
Each network service supports multi-device management with conflict detection:
| Service | Capabilities |
|---|---|
| DHCP | IPv4 & IPv6 servers, address pools, static leases, live lease table |
| DNS | Upstream servers, static records (A/AAAA/CNAME/MX/NS/PTR/TXT/SRV), cache flush, DoH |
| NTP | Server (broadcast/manycast), client (unicast/multicast), sync status |
| WireGuard | Interface management, peer configuration, public keys, RX/TX stats |
| Logging | Syslog actions and routing rules, single-device or push-to-all with per-entry coverage |
| NetFlow | One-toggle Traffic Flow export per device, pointed at the built-in collector |
| Discovery & SNMP | Fleet-wide LLDP toggle and SNMP v1/v2c/v3, scoped to all devices, routers, or switches |
Visibility and troubleshooting
- Network topology — auto-discovered from LLDP, CDP and MNDP, with LLDP treated as ground truth and lower-priority protocols suppressed for the same neighbour. Bidirectional pairs merge into one edge with both port names. Neighbours are resolved to managed devices by MAC before IP, because a MAC is unique fleet-wide and an address is not
- Manual links — drag between any two devices to record a connection discovery cannot see; stored persistently and drawn as purple dashed edges
- Orphan detection — devices with no known connections are grouped with a prompt to link them
- Traffic analytics — a built-in NetFlow v9/IPFIX collector on UDP 2055, per-client attribution, application breakdown (HTTPS, QUIC, DNS, SSH, email, WireGuard…), top talkers over 1h → 30d, automatic deduplication when a flow crosses two managed routers, and NAT-tolerant ingest for routers exporting from behind NAT or a VPN subnet router
- Configuration history — snapshots of each device's full
/export, deduplicated by content hash so only real changes are stored, with side-by-side line diffs, a change summary, and one-click rollback through the proven restore path - Audit log — every authenticated write recorded with user, timestamp, method, path, entity, IP and response status
- Global search — devices, clients and events from the top bar
- Per-device tools:
| Tool | Description |
|---|---|
| Ping | ICMP reachability with RTT and loss |
| Traceroute | Hop-by-hop path trace |
| IP Scan | ARP sweep of a subnet to find live hosts |
| Wake-on-LAN | Magic packet sent from the MikroTik device |
| Packet Capture | RouterOS sniffer for 5–60s, downloaded as a .pcap for Wireshark (requires SSH) |
| Bandwidth Test | Throughput between two devices; the target's test server is enabled and disabled automatically |
Fleet operations
- Firmware orchestration — fleet version overview, live update checks, and MikroTik's official release notes in-app. Staged rollouts run in waves (wave 1 = canary) through a verified pipeline per device: pre-upgrade backup → download and confirm the image landed → reboot → prove it restarted and came back healthy on the new version → next. Halt on failure stops the rollout if any device fails, and a device that comes back on the old version counts as a failure. Devices in a wave upgrade one at a time by default; raise at once to run them concurrently when a fleet of five-minute reboots would otherwise take all evening — waves stay sequential either way, so the canary still gates the rest. Rollouts can be scheduled, and cancelling never interrupts an in-flight flash
- Backups — on demand or on a daily/weekly/monthly schedule for all online devices. Filter by device, type and date range; read one in place instead of downloading it; and compare any two as a diff — they are
/exporttext, not opaque blobs. Config snapshots and their restorable.rscare one artifact: delete either and both go, so they never drift apart, and a bulk delete says how many snapshots it will take with it - Configuration templates — reusable sets (DNS, NTP, syslog) pushed to many devices with per-device result reporting
- Bulk device add — "Try All" on discovered devices runs as a server-side job that survives a closed browser tab, with live progress and cancel
- Device organisation — colour-coded tags, rack location, physical address with map support, and per-device notes
- Sites — group devices into sites and see each one's devices, clients, events, topology and dashboards on their own. A selector under the logo switches between them, and an all-sites view pins every site on a world map. Single-network installs are collected into one "Default Site" and never meet the concept. Polling stays fleet-wide — sites are a view over the data, not a change to how it is gathered. See the sites guide
Access control and platform
- Roles — Admin, Operator (read/write), Viewer (read-only)
- Two-factor authentication — per-user TOTP via QR code, required at login once enabled
- Single sign-on (OIDC) — any standards-compliant provider (Entra ID, Okta, Google Workspace, Keycloak, Authentik…), with group-to-role mapping, auto-provisioning, and local login retained as break-glass. Configured entirely in the UI — see the SSO guide
- Scoped API tokens —
mtm_…tokens with read or write scope and optional expiry, shown once and stored only as a SHA-256 hash. No token can perform admin actions or manage other tokens - Outbound webhooks — subscribe to any of twelve events, delivered as HMAC-SHA256-signed JSON POSTs through the same pipeline that respects alert rules, cooldowns and maintenance windows
- Scheduled email reports — daily, weekly or monthly HTML fleet summaries to any recipient list
- Credential presets — shared device credentials, optionally restricted to admins only
- Encryption at rest — device passwords encrypted with AES-256-GCM under a self-healing key
- TLS — self-signed certificate generated on first run, replaceable via the Settings UI; nginx terminates TLS and redirects HTTP
Documentation
Deeper reference material lives at the documentation site, published per release so it matches the version you are running. The same pages are in docs/ if you would rather read them here.
so it moves when behaviour does.
| Document | Covers |
|---|---|
| Configuration | Environment variables, secret management, key rotation, TLS |
| Change Guard and Config Health | How the safety system works, its settings, and what to expect when it fires |
| Sites | Grouping devices into sites, switching between them, and what each site scopes |
| Firmware rollouts | Waves, the upgrade pipeline, and what upgrading several at once costs |
| Backups | Finding, reading, comparing and deleting configuration backups |
| Alerting | Rules, channels, ntfy priority mapping, webhooks, maintenance windows |
| Single sign-on (OIDC) | Identity provider setup, group-to-role mapping, break-glass behaviour |
| CAPsMAN | How centrally provisioned APs are modelled, and current limits |
| API and automation | Scoped tokens, preflighting a change, driving the platform from scripts |
| Architecture | Data stores, polling model, and where the interesting logic lives |
Tech stack
| Layer | Technology |
|---|---|
| Frontend | React 18, TypeScript, Vite, Tailwind CSS |
| State / data | TanStack Query v5, React Router v6, Zustand |
| Charts / topology / maps | Recharts, @xyflow/react, Leaflet |
| Backend | Node.js, Express, TypeScript |
| Data stores | PostgreSQL 15, InfluxDB 2.7, Redis 7 + BullMQ |
| Real-time | Socket.IO |
| Device comms | RouterOS API (8728 / 8729), SSH2 |
| Proxy / container | nginx, Docker Compose |
Contributing
Contributions are welcome. Please open an issue before submitting a pull request so we can discuss the approach — several recent features started as exactly that conversation.
- Fork the repository
- Create a branch:
git checkout -b feature/your-feature - Commit your changes
- Push and open a pull request
License
Licensed under the GNU Affero General Public License v3.0 — see LICENSE for the full text.
- You are free to use, modify and distribute this software.
- If you run a modified version as a network service, you must make your modified source available to users of that service under the same license.
- Distributed copies and derivatives must also carry AGPLv3.
This license keeps improvements — including those deployed as a service — open and available.
AI assistance
This project was designed and built with the help of Claude by Anthropic. AI assistance was used throughout: architecture, backend services, frontend components, the CI/CD pipeline, security configuration and unit tests.
We believe in being transparent about how software is made. The code is reviewed and tested with AI assistance and maintained by the project authors.
Disclaimer
Not affiliated with or endorsed by MikroTik. MikroTik and RouterOS are trademarks of SIA MikroTīkls. Use at your own risk, and always test configuration changes outside production first.