Documentation

SOVA documentation

From the one-line install to a whole company on one routed network.

Quick install

A hub in minutes, from one command.

Run it as root on a fresh Ubuntu 24.04 or 26.04 VPS with a public IP.

Read the full installation guide
Ubuntu — bash
curl -fsSL https://sova.securytik.com/install.sh | sudo bash

Getting started

Overview

SOVA turns one Ubuntu VPS into the hub of a company network. Every branch office runs a MikroTik router (RouterOS 7) that dials out to the hub — so branches behind NAT, CGNAT or a 4G modem work without a public IP — and every remote employee connects to the same hub. Everything can then reach everything else by its real address, as if all the offices were in one building.

PieceWhat it does
Hub (your VPS)Terminates every tunnel, routes between sites with OSPF, enforces policies, answers internal DNS, runs the panel.
SiteOne branch router. Gets its own address block, a WireGuard tunnel and an L2TP/IPsec backup tunnel to the hub.
Network typeA purpose — servers, users, printers, cameras… — that sits at the same third octet at every site.
Remote userAn employee device on WireGuard or native L2TP/IPsec, with access limited to the sites and networks you choose.
ManagerHealth and control of the branch MikroTiks through the tunnels: updates, time, DNS, app filter, bulk actions.

You never write router configuration by hand: SOVA generates a paste-ready script per branch, and adding a new branch later does not require touching the existing ones.

Getting started

How SOVA works

  1. The hub owns the plan. Every address in the fleet is derived from a site number and a network type, so nothing is typed twice and two sites can never overlap.
  2. Branches dial out. Each branch router opens a WireGuard tunnel to the hub (UDP, its own port per site) and, as backup, an L2TP/IPsec tunnel. Outbound connections pass any NAT.
  3. OSPF shares the routes. The branch advertises its own networks; the hub learns them and advertises the whole fleet back. A new branch is learned by every other branch automatically.
  4. The hub is the checkpoint. All site-to-site and remote traffic passes through the hub, so policies, remote-user access and route validation are enforced in one place — a branch cannot widen its own access or claim another branch's addresses.
  5. Local internet stays local. Only company traffic crosses the tunnels. Branches keep using their own internet; remote users choose split or full tunnel.

Getting started

Requirements

Hub

  • A VPS with Ubuntu 24.04 or 26.04 (amd64), freshly installed, root access.
  • A public IPv4 address on the VPS (not behind NAT) — branches and users dial it.
  • 1 vCPU / 1 GB RAM for up to ~10 sites; 2 vCPU / 2 GB for dozens of sites and users; 10 GB disk.
  • If your provider has an external firewall, allow inbound: TCP 80 (panel), UDP 51001–51254 (site N uses 51000+N), UDP 51820 (remote users), UDP 500, 4500 (IPsec) and your SSH port. SOVA configures the VPS firewall itself.

Branches

  • MikroTik with RouterOS 7 (any model, CHR included) and working internet.
  • A free port for each network you want SOVA to build (see Creating sites).
  • The branch must not already use addresses inside its future block 10.N.0.0/16 or 172.31.0.0/16.

Remote users

  • WireGuard app (Windows, macOS, Linux, Android, iOS), or the L2TP/IPsec client built into Windows, macOS, iOS and Android.

Getting started

Installation

On the fresh VPS, as root:

curl -fsSL https://sova.securytik.com/install.sh | sudo bash

If github.com is blocked in your country, use the Cloudflare-hosted mirror instead — same installer, nothing fetched from GitHub. (The standard command above also detects a blocked GitHub and switches to this mirror by itself.)

curl -fsSL https://dl.securytik.com/sova-install.sh | sudo bash

The bootstrap downloads the latest signed release from GitHub (github.com/mhdhaidarah/sova) — or from the dl.securytik.com mirror when GitHub is blocked — verifies its checksum and runs the installer inside it. The installer:

  1. installs WireGuard, FRR (OSPF), strongSwan + xl2tpd (L2TP/IPsec), nftables, Unbound (DNS), PostgreSQL and nginx;
  2. creates the database and the sova service account, and the SOVA services;
  3. detects the public IP and serves the panel on it (http://<server-ip>/admin/);
  4. prints the panel address and a random first password.

Re-running the same command on an existing hub upgrades it in place and keeps every setting and site. The full log is in /var/log/sova-install.log.

Services it installs

ServiceRole
sova-apiThe web panel (behind nginx).
sova-agentApplies the hub configuration (tunnels, OSPF, firewall, DNS) whenever something changes.
sova-monitorMeasures tunnels, routes, remote users and devices; raises alerts; sends webhooks.
sova-health-watch (timer)Restarts any SOVA or network service that stopped or stopped working.
sova-updater, sova-license-enforcer (timers)Daily update check, hourly licence check.

Getting started

Set up your server

No VPS yet? We set one up for you. Order it from your SOVA dashboard, pay, and receive a ready SOVA hub on your own .com domain, with your name on it. There is nothing to install.

  • A cloud server and a .com domain, in three sizes: 6 cores, 12 GB RAM, 200 GB SSD for $100 a year; 8 cores, 24 GB RAM, 300 GB SSD for $200; 12 cores, 48 GB RAM, 400 GB SSD for $300. You pick the size in the first step.
  • HTTPS on your domain; the server's own address is never published.
  • SOVA installed and branded with your company name, with your licence linked and active at the first login.
  • An e-mail with your panel address and a strong admin password.
  1. Sign in at sova.securytik.com and open Dashboard → Get a server.
  2. Choose a server name and the number of users, then search for your .com domain or pick one of the suggestions.
  3. Pay with any method shown at checkout. The order waits two hours for the payment; after that it is cancelled and you place a new one.
  4. After the payment everything is automatic. When the hub is ready you receive the e-mail with your address and login.

Your own SSH login (password and/or public keys) is managed under Dashboard → Servers, with a step-by-step SSH guide. Renew at least one day before the end of the year (renewal opens seven days before); if it is not renewed, the server, the domain and the data on it end.

Getting started

First login

Panelhttp://<server-ip>/admin/ · Usernameadmin · Passwordprinted by the installer

The first password is also stored in /etc/sova/first-admin-password (root only). For HTTPS on your own domain, use Cloudflare Tunnel. After signing in:

  1. change the password under Profile (top right) and add your email for password recovery;
  2. check System → Settings → Hub — the public address must be what branches can reach;
  3. activate the install with your SecuryTik account under System → License — until then only that page opens.

Getting started

Hub settings

System → Settings:

SettingMeaning
Public addressIP or DNS name the branches and remote users dial. Changing it means re-pasting every branch script and re-sending user profiles. Detect checks the hub's real public address and warns when the saved value does not lead back to this hub.
Internal DNS zoneThe zone the hub answers for, e.g. corp or office.local (see Internal DNS).
Fallback DNSSecond DNS server handed out by branch DHCP, used if the tunnel is down.
Time zoneGeneral tab: the time zone the whole panel shows times in.
Alert channelsMoved to System → Notifications (Telegram, email, WhatsApp), with per-event rules and an outbox.

The page also shows the hub's WireGuard public key, the port plan and the result of the last configuration apply.

Building the network

Address plan

RangeUse
10.N.0.0/16Site N (1–254).
10.N.T.0/24Network type T at site N; the router is 10.N.T.1.
10.255.0.0/17Remote users — Corporate profile (one fixed address per user).
10.255.128.0/17Remote users — Full passthrough profile.
172.31.0.1The hub itself: internal DNS and OSPF router ID.
172.31.N.0/30, 172.31.N.6Site N's WireGuard and L2TP tunnel addresses.

Example: with type Servers = 1 and Printers = 3, branch 2's file server lives in 10.2.1.0/24 and branch 5's printers in 10.5.3.0/24. The hub infrastructure range is deliberately outside 10.0.0.0/8 because many ISPs and home routers use 10.0.x.x on the WAN side.

Existing branch LANs are re-addressed into the site's block. If a branch must keep an old LAN for a while, leave it on its port and let SOVA build the new networks on other ports; move devices over, then retire the old LAN.

Building the network

Network types

Network → Network types. A type fixes the third octet across the whole fleet. SOVA ships with Servers (1), Users (2), Printers (3), Cameras (4), Voice (5) and Guests (6); add your own (1–254) with a name, a short name (used in DNS names) and a colour.

Because a type means the same octet everywhere, one policy or one remote-user rule can say "printers at every site" instead of listing addresses. A type in use by a site cannot be deleted.

Building the network

Creating sites

  1. Network → Sites → + Site. Enter the name, keep or change the site number (it becomes the second octet), the management port (the port you manage the router from — SOVA never builds a network on it), and whether to add the L2TP/IPsec backup and let the Manager control the router.
  2. On the site page, add networks: pick a type, the physical port (e.g. ether2) and the DHCP range. One port per network — no VLANs needed. The port is taken out of any bridge.
  3. Copy the RouterOS script or press Make a one-line install link and paste the one-liner into the router (see The branch script).

The site page shows the live state: WireGuard and L2TP up/down, which path OSPF is using, latency, loss, the routes the branch advertises (and any the hub rejected) and a 24-hour latency chart.

Rotate keys issues a new WireGuard key and L2TP password (the branch goes offline until the new script is pasted). Delete site closes its tunnels and removes its routes everywhere.

Changing a site later

On the site page: Edit site (name, enabled, L2TP backup, managed, notes), + Network, and the ⋮ / right-click menu on each network to change its port or DHCP range or remove it. The Sites list itself has row selection with bulk enable / disable / delete, a column chooser, search, filters and export.

Internet through the hub (a switch on each network, off by default) sends that network's Internet traffic out through the hub's public address instead of the branch's own line. Traffic to other sites keeps its path. If the hub cannot be reached, the network uses the branch line until it can.

Building the network

The branch script

Each site shows one command. Paste it into the router's terminal (WinBox → New Terminal, or SSH):

/tool fetch url="http://<hub>/s/<token>.rsc" dst-path=sova.rsc; :delay 2s; /import sova.rsc; /file remove sova.rsc

The link works once, for 24 hours, and only for that site; New link on the site page makes another. The script:

  • checks first — it stops without changing anything if the router already uses the site's block or 172.31.0.0/16, and warns about other 10.x addresses;
  • builds the networks (address, pool, DHCP server with the hub as DNS);
  • creates the WireGuard interface and peer, the L2TP/IPsec client, OSPF with import/export filters;
  • adds firewall rules that accept fleet traffic for this site's block and skip NAT for it;
  • adds the Manager login (usable only from the hub) when the site is managed;
  • turns on the NTP client (time.cloudflare.com, pool.ntp.org), so the router's clock is right;
  • never touches the WAN, the default route or anything not tagged SOVA.

Paste it again at any time — after adding a network, rotating keys or changing the hub address. It removes its own SOVA entries and re-creates them. To remove SOVA from a router, delete the objects whose comment starts with SOVA.

Apply to router

For a managed site, Apply to router on the site page sends the current script over the tunnel: the router downloads it from the hub and imports it in the background. The tunnel blinks for a few seconds. Paste the script by hand once first, so the Manager login exists.

Building the network

Routing & failover

  • OSPF runs on every tunnel (hello 1 s, dead 4 s). WireGuard has cost 10, L2TP cost 100, so WireGuard is always preferred.
  • Failover: if WireGuard stops (blocked UDP, ISP issue), traffic moves to L2TP/IPsec within about 4 seconds and returns automatically. The dashboard and alerts show when a site is on the backup path. Branches set up before SOVA 2.0 keep the older 20-second timers until their router is synced or the script is pasted again.
  • Route checks: the hub installs a branch's routes only when they fall inside that branch's own block and arrive over its own tunnels. Anything else is rejected and raises an advertises foreign routes alert — a misconfigured branch cannot hijack another branch's traffic.
  • The hub advertises summaries of the whole fleet, so every branch always sends company traffic to the hub.

Network → Routes (under Overview) lists what each site advertises, what was rejected and the hub's OSPF neighbours.

Networks made on the router

Each site has a Share networks made on the router toggle (on by default). On: any interface the branch admin adds by hand inside the site's block is announced to every site. Off: only the networks SOVA built are shared. Networks outside the site's block are never shared — they would clash between branches; give such an interface an address in the site's block instead.

Building the network

Direct links between sites

Since SOVA 2.0, two sites talk to each other directly — router to router over the Internet — whenever their connections allow it, instead of through the hub. The hub stays connected to every site and carries the traffic whenever a direct link is not possible or stops working, and the traffic returns to the direct link by itself.

  • How: every site gets a second WireGuard interface, sova-mesh, with one peer per other site. The hub learns each site's public address and port and hands them to both routers at once, so both NATs open (UDP hole punching).
  • Which pairs: the hub checks each site's NAT on two ports. Open (no port translation) and Cone sites link directly; a Symmetric NAT (common behind carrier-grade NAT) links only to open sites and otherwise uses the hub. The site page shows its type.
  • Routing: OSPF runs over the direct links too. A direct link costs 15, the path through the hub 20, and a path through a third branch 30 — so traffic goes direct first, then through the hub, and never through another branch while the hub path exists. A broken direct link is left in about 4–6 seconds and comes back in about 6.
  • Stability: a pair that goes up and down too often is held on the hub for a while, and a pair that worked directly but has used the hub for more than 10 minutes raises an alert. Both limits are in Settings → Monitoring.
  • On the map: a solid line is a working direct link, a dashed line a direct link that is down (the hub carries it). Hub links and Direct links next to Flow show or hide either.

Turn it off per site with Direct links to other sites in Edit site; everything then goes through the hub. Two sites behind the same public address need hairpin NAT on their shared router to link directly.

Reachable at. A site behind carrier-grade NAT, with no address the hub can learn by itself, can still be given one by hand in Edit site — its fixed public IP, or a UDP port its own modem forwards to the router's port 51900. Other sites then dial that address directly instead of going through the hub. Leave it empty and the hub keeps learning the address on its own, as before.

Building the network

Remote users

Remote Access → Remote users → + Remote user.

FieldMeaning
ProfileCorporate: only company networks go through the VPN. Full passthrough: all traffic leaves through the hub (the user browses with the hub's IP).
ProtocolsWireGuard, L2TP/IPsec, OpenVPN (for IP phones and devices without WireGuard), or any mix.
DNSWhat the profile gives the device as DNS: the hub (internal names work), none (the device keeps its own DNS — good for phones that stay connected all day) or custom servers.
ExpiresThe user is cut off automatically on that date.
AccessWhat the user can reach: the whole fleet, a whole site, one network type everywhere, or one network at one site. Without any rule the user connects but reaches nothing.

New key & password invalidates the old profile. Disabling or deleting a user disconnects them immediately.

Groups (Sales, IT, Contractors…) gather remote users so one policy covers them all: create them under Remote users → Groups, tick users and choose Set group. A policy can also name one single remote user.

Building the network

Connecting clients

WireGuard (recommended)

Install the official WireGuard app, then scan the QR code on the user's page (phones) or import the downloaded .conf (computers). Works from any network, including behind other NATs.

L2TP/IPsec (built in)

  • Server: the hub's public address · Type: L2TP/IPsec with pre-shared key · the key, username and password from the user's page.
  • Windows behind NAT needs, once, the registry value HKLM\SYSTEM\CurrentControlSet\Services\PolicyAgent\AssumeUDPEncapsulationContextOnSendRule = 2 (DWORD) and a reboot.
  • Corporate profile on Windows: untick "Use default gateway on remote network" and add routes to 10.0.0.0/8 and 172.31.0.0/16 via the VPN.
  • Two L2TP/IPsec users behind the same public IP interfere with each other — give them WireGuard.

Remote users get the hub (172.31.0.1) as DNS, so internal names work — unless the user's DNS is set to none or to custom servers.

OpenVPN

For devices that have no WireGuard — most IP desk phones. Turn on OpenVPN for the user; the user's page then shows its username and password (with copy buttons) and offers the .ovpn profile two ways — with password (connects at once) or without password (OpenVPN Connect asks for them) — plus ca.crt, client.crt and client.key for desk phones. UDP port 1194 on the hub opens only while at least one user has OpenVPN. New key & password issues a new certificate and the old one stops working at once. See IP phones & PBX.

One-line installers (Windows and Ubuntu)

The user's page has a Set up a computer card: pick the connection (WireGuard or L2TP/IPsec) and the device, then one button per job — Install, Connect, Disconnect, Remove — each copying a one-line script to paste into PowerShell (Windows) or a terminal (Ubuntu). Connect and Disconnect both hold after the computer restarts. Pasting the installer twice is safe: a new paste replaces any older SOVA tunnel on that PC, even one made for another Windows user, so a PC never ends up with two. Phones and Macs still get the QR code, the .conf file and the L2TP details.

Handing out the settings

From the Remote users list (⋮ or right-click) or the user's page: Copy WireGuard config, Export WireGuard (.conf), Copy L2TP details, Export L2TP details (.txt).

Building the network

IP phones & PBX

SOVA carries a company phone system with nothing opened on the Internet: the PBX sits on a private network at one branch, and every other branch, desk phone and mobile reaches it by its real address through the tunnels.

NeedHow with SOVA
Desk phones at the branchesPlug them into a network of the branch router; they register on the PBX's private address (for example 10.1.1.53).
A small shop or kioskA small MikroTik (hAP ax lite, or LTE6 with a 4G SIM where there is no line) added as a site; the phone and the till plug into it.
Two PBXs, two citiesA SIP peer trunk between them over the tunnel — no public SIP port at either end.
Extensions on mobilesA remote user per mobile (Corporate, WireGuard, always-on VPN) and the vendor's softphone app pointed at the PBX's private address — no vendor cloud plan needed for remote extensions.
A desk phone at home or a site without a routerA remote user with OpenVPN; upload the three files on the phone's OpenVPN page.
Managing the PBX from scriptsThe API's POST /proxy sends HTTP(S) requests to the PBX's own API through the hub — see REST API.

Setting up a desk phone over OpenVPN

  1. Create a remote user (Corporate) with OpenVPN on, and give it access to the PBX's network.
  2. On the phone's web page, open the OpenVPN settings: enable OpenVPN, server = the hub's public address, port 1194, UDP, encryption AES-256-CBC, LZO compression off, and the username and password from the user's page.
  3. Upload ca.crt, client.crt and client.key from the user's page, save and reboot the phone.
  4. Register the extension on the PBX's private address. The user's page shows the phone online with its traffic.

Tips: give always-on phones DNS none so their normal browsing keeps the mobile network's DNS; use the PBX's own ring groups per branch (for example 1xxx, 2xxx, 3xxx) and one public number on the main PBX.

Building the network

Policies

By default every site reaches every site. Network → Policies adds allow/deny rules between any networks of the fleet, enforced on the hub:

  • From / To: any network inside the fleet — a site block, one network, all remote users (10.255.0.0/16)…
  • Protocol and port (e.g. TCP 445), and whether the rule applies in both directions.
  • Order: rules are checked top to bottom; the first match wins. A deny also cuts connections that are already open.

Example: deny 10.0.0.0/8 → cameras 10.1.4.0/24, then allow the security office 10.2.2.0/24 → cameras above it. Remote users are governed by their own access rules; policies apply on top.

Several sources and destinations

In the rule pop-up, choose a kind (whole fleet, site, site network, remote users, IP / network), then Add to the list to put several of them on the same side of one rule.

Blocking apps and websites

Choose App / website as the destination and tick the apps (TikTok, Instagram, games, streaming…). Branch internet does not pass through the hub, so these rules are pushed to the source sites' routers — managed sites only. SOVA switches the apps on in each router's Website & app filter and adds drop rules tagged SovaPolicy:. The same blocks can be made from a router's Websites & Apps tab in the Manager (Blocked for networks). Use Push app rules to routers after re-pasting a script or adding a site.

Building the network

Internal DNS

The hub answers for the internal zone (Settings) on 172.31.0.1 and forwards everything else to public DNS. Branch DHCP and remote users use it automatically. Names created for you:

NamePoints to
hub.<zone>the hub
router.<site>.<zone>the branch router's tunnel address
gw-<type>.<site>.<zone>the gateway of each network

Add your own under DNS & Devices → DNS: type the device name, pick its site, and the zone is added — erp at site hq becomes erp.hq.corp. The address must be inside that site.

Building the network

Devices & static leases

DNS & Devices → Devices lists every device behind your managed branch routers — their DHCP leases and ARP tables, read about every minute — with search, filters and grouping (site, network, lease, online). Right-click a device to:

  • Make its lease static, or move a static lease to a new address: the device takes it when it next renews or reconnects.
  • Give it a DNS name that follows it: when its static lease moves, the name moves too.
  • Forward a port to it, or add it to Wake on LAN.

Networks → Show devices opens the list filtered to one network type across every site.

Building the network

Port forwards

A port forward opens a port on the hub's public address and sends what arrives to a device in a site — by address, or by DNS name (the rule follows the name). The device sees the caller's real address: the branch router sends the replies back through SOVA. Optional Allow from limits who may connect; ports the hub itself uses are refused.

Each rule counts its connections (Hits), and Hits lists every forwarded connection — when, from which address — with search and export. Do not forward services with weak or default passwords to the whole Internet.

Building the network

Wake on LAN

Devices added from the Devices list can be woken from DNS & Devices → Wake on LAN: the site's own router sends the wake-up packet. With Wake on traffic (needs a static lease) the router wakes the device by itself when a connection it routes heads for it — from the hub, another site, a remote user or a port forward. Devices on the same LAN reach it directly, without the router, so they do not wake it.

Operating

Dashboard

The dashboard is made of blocks you arrange: drag a block by its header, resize it from the corner, remove it with ×, add more with Add block and save the arrangement — or save several named dashboards, which appear under Dashboard in the menu. The default top row is a chart of connected sites and Remote Access users online over time, beside System Health (version, CPU and RAM with history, disk, uptime, services, last monitor pass, licence, last hub apply). Other blocks: open alerts, tunnel latency & loss, tunnel traffic, sites, routes, remote users, managed devices, device health and recent admin actions. Charts pick their sites and time window from their own header.

Announcements from SecuryTik — news and product messages — appear as a pop-up on the dashboard when one arrives with the normal update check (news shows once, other messages at most once a day). Nothing about the hub is sent to see them.

Operating

Network map & tools

Network → Network map draws the fleet as it is right now: the Internet, the hub, every site with its live tunnel (WireGuard, L2TP backup or down), its transit addresses and networks, and Remote Access. Drag boxes to arrange it — the arrangement is remembered — and save arrangements as bookmarks. Layers switch interfaces, IP addresses, networks, CPU and the traffic flow on and off. Right-click a site to open it, ping or traceroute it, or manage its router.

Ping, Trace and Speed run on the map itself, in both Diagram and Map (GeoMap) modes: click the hub or a site, then the target. Echoes and traceroute hops are drawn as they happen; the result opens in a window or stays on the line. Speed measures site to site (both directions), site to hub and site to Internet.

Tools → Network runs ping, traceroute and DNS lookup on the hub or on any managed branch router: a target that answers from the hub but not from a branch points at that branch's path.

Operating

Performance

Network → Performance shows the last speed test and ping between every pair — sites, the hub and the Internet — coloured by speed. Click a cell to measure again. Below it, every result is kept with what may have held it back: a router's CPU at its limit, one direction far slower than the other (usually a site's upload), lost packets or high latency.

Check performance now measures ping and speed for every pair; several pairs run at once, but a site is never in two tests at the same time. In Settings → Monitoring it can run by itself every few hours; it is off by default, because each speed test fills the line for a few seconds.

Speed to the hub is measured both ways. Speed to the Internet is download only: a router cannot upload to an Internet speed server.

Operating

Monitoring & alerts

Every 30 seconds the hub checks each tunnel (handshake, OSPF neighbour, ping), each remote user and each managed device, and keeps 30 days of history (Overview → Monitoring).

AlertWhen
Site down (critical)Neither tunnel carries traffic for 1.5 minutes.
On the backup path (warning)WireGuard is down and the site runs over L2TP/IPsec.
Advertises foreign routes (warning)A branch announces addresses outside its block.
Device downA managed router stops answering.
Hub configuration apply failedA change could not be applied to the hub.
The license ends soon30, 14, 7, 3 and 1 days before the plan ends, with the end date.

Alerts resolve by themselves and are sent (opened and resolved) to Telegram, email and WhatsApp as configured in Settings. The bell at the top of every page shows what needs attention.

Handling alerts

Overview → Alerts: open alerts are under Needs attention. I am on it moves one to Being handled (still open, no longer counted in the bell); Quiet 1h / Quiet a day hide it for a while. The history below is searchable and exportable.

Notification center

System → Notifications: set up Telegram, email and WhatsApp Cloud API under Channels, choose which events go where under Admin alerts, and see every message sent or failed in the Outbox.

Operating

Managing branch routers

Every site page has two tabs: Site (addresses, tunnels, script) and Manage (the router itself, controlled through the tunnel). For a site SOVA does not manage yet, the Manage tab offers Manage this router: paste the install command once, and from then on changes deploy over the tunnel. Sites → All routers lists every router at once for bulk actions, and routers added by address. On each router:

  • Overview: model, RouterOS version, CPU, memory, uptime, interfaces with traffic, health history.
  • General: identity, time zone, NTP and DNS servers, reboot.
  • Website & app filter: block websites and apps by category (see next topic).
  • Update: check, download and install RouterOS updates.
  • Tools: ping and traceroute from the router, ping from the hub.

Bulk actions: tick several devices and set the time zone, NTP or DNS servers, check or install updates, take a backup, reboot or run a ping — with a result per device.

The router login the branch script creates (sova-mgr) accepts connections only from the hub's tunnel range.

Config delivery retries by itself. A router that could not be reached is tried again after 1, 2, 5, 10 and 30 minutes — six tries in all — and the delivery page shows when the next try is due. Retry by hand, or fall back to copying the script, at any time.

Traffic history

The router's Overview tab shows Interface traffic (download / upload per interface) and App traffic (per app and website, with totals) for windows from 10 minutes to 7 days. Samples are taken every minute and kept 30 days. App traffic counts the apps switched on in Websites & Apps.

Live changes. On a managed router, saving a network (port, DHCP, range), a policy or the hub's DNS servers updates the router at once, record by record over its API: the script is not re-imported and the tunnel never drops. Policies between two networks of the same site — traffic that never reaches the hub — are enforced on that site's router, in the panel's order. The site page shows the last router sync and a Sync now button, and the hub re-checks every managed router every 10 minutes.

Operating

Router security

On a managed site, Manage → Tunnel tab → Router security secures the branch router with SecuryTik's lab-tested MikroTik rules — the same engine as SAMM's Auto Security. One click picks a level; off by default.

LevelWhat it does
NormalConnection tracking, an ICMP policy, bogon drop, no open DNS resolver, login brute-force protection, and service hardening (Telnet, FTP and HTTP off; SSH with strong crypto only). Safe on any router.
AdvancedEverything in Normal, plus: unsolicited Internet traffic dropped at the router and into the LAN (port forwards still work), DDoS, SYN-flood and port-scan protection, DHCP starvation and rogue-DHCP protection, and management allowed only from the LAN and the fleet.

Inter-site traffic and the Internet keep working at both levels — SOVA's own fleet rules always sit above every security rule — and neighbor discovery keeps RouterOS's default behaviour.

Manage → Firewall → Security shows every protection as its own switch: turn one off or on by itself and the level reads Custom. The manual Filter Rules tab is untouched. Every change takes a backup of the router first, checks that SOVA can still log in, and puts the backup back by itself if not.

Operating

Website & app filter

On a device's Website & app filter tab, tick the apps and categories to block (social networks, video, games, messaging…). SOVA pushes DNS-based blocking to the router from SecuryTik's maintained catalogue and keeps it updated when you press Update. Its rules are tagged separately, so re-pasting the branch script does not remove them; Remove takes them all off.

Operating

Traffic

Network → Traffic (after Monitoring) draws live traffic for the whole fleet on one chart that keeps moving on its own: each site to the hub, between sites over their direct links, each site to the Internet, the hub to the Internet, and remote access.

  • Read every 3, 5, 10, 15 or 30 seconds (10 by default), or down to every second for a closer look — the panel warns that a 1-second read costs more CPU on the hub.
  • Show the last 1 to 10 minutes; filter by site, direction and type; keep only the busiest lines (Top 5, Top 10 or All).
  • Click a line to show only it, Ctrl + click to add more. Bookmarks remember your picks, including the selected lines, so a saved view comes back exactly as you left it.
  • Export the chart and the list as a landscape PNG.

Operating

Search, filters & export

Sites, remote users, devices, alerts and the audit log share the same list tools: type to search, add filters from the menu, group by any field, build custom rules (e.g. latency above 80 ms), sort by clicking a column, save a search as a favourite (private or shared) and export to CSV or Excel. The search button at the top of every page (Ctrl K) finds sites, users, DNS names and pages.

Operating

Admins & roles

System → Admins adds administrators. A role sets, per area (sites, remote access, policies, DNS, monitoring, Manager, settings…), whether the admin cannot see it, can view it or can edit it. SOVA ships with Super Admin (everything), Manager (runs the network) and Monitor (read-only); you can create your own. Only a Super Admin manages admins, roles and API tokens.

Branches. Admins are arranged in a tree. A branch owns its sites and remote users: its admins see and manage only those (and the branches below it), while admins whose role sees everything see the whole hub and choose the branch when they create or edit a site or a remote user.

Forgot password: the sign-in page sends a one-time code to the admin's email (configure SMTP in /etc/sova/smtp.env).

Each admin picks their own panel language and theme under Profile. A superadmin can edit any translation under System → Translations (applied live), add a language, or export a language to Excel and import it back.

Operating

Reset admin password

Locked out and no email recovery? Run the standalone reset tool on the hub, as root. It sets a new password for a Super Admin account straight in the database and re-enables the account — no working login needed, and no site, tunnel or user data is touched.

curl -fsSL -o sova-reset-admin-password.sh \
  https://raw.githubusercontent.com/mhdhaidarah/sova/main/tools/sova-reset-admin-password.sh
sudo bash sova-reset-admin-password.sh --list   # list the superadmin accounts
sudo bash sova-reset-admin-password.sh          # reset (asks for the new password twice)

With several Super Admins, pick one with --user NAME. The password is typed hidden, never on the command line, and stored hashed exactly like the panel does. The tool also ships inside every install at /opt/sova/tools/.

Operating

Audit log

System → Audit Log records every change with who made it, when, from which address and what changed — from the panel and from the API — plus sign-ins and failed sign-ins.

Operating

Backup & restore

Tools → Backup & Restore takes a full backup of SOVA's database (sites, keys, users, policies, settings, admins), schedules a daily automatic backup with retention, downloads and uploads backups, and restores one. After a restore the hub re-applies itself within seconds.

Moving to a new VPS no longer needs a backup by hand: see Move to a new server.

Operating

Move to a new server

Settings → Migration moves a whole hub to another VPS — sites, networks, keys, users, policies, settings and the licence — from two browser windows, with no command line. Your routers keep working the whole time and each one is switched to the new server separately.

  1. New server: install SOVA on a fresh Ubuntu VPS (the same one-line installer).
  2. Old hub → Move this hub away → Prepare. SOVA packs everything and shows the old hub's address and a one-time code (valid 24 hours). Changes are locked on the old panel from now on, because the package is a snapshot.
  3. New hub → Receive a hub here: enter the old hub's address and the code. The new server fetches the package, restores it and starts; its panel then asks you to sign in with the old hub's administrator account.
  4. Old hub → Move all sites. Every managed router is switched to the new server by a script that undoes itself if the router cannot reach the new hub, so a wrong address cannot cut a branch off. A router SOVA does not manage shows the new hub's script to paste. Each site shows Moved, Stayed here (and why, with a Retry) or Paste script.

The licence

The licence moves with the hub: the new server gets its own device record with the same plan and expiry, and the licence server ends the old hub's licence the moment its last site has moved. Nothing is paid twice and nothing needs re-activating.

What happens to the old server

When every site has moved the old hub's page says so and plans its own reset: seven days later it erases itself — sites, users, settings and every backup — and comes back as a fresh SOVA install. The administrator accounts keep the same usernames and passwords, so the server can be used as a new hub straight away. To do it at once press Reset now and type RESET.

Remote users download their profile again from the new hub, because the profile carries the hub's address.

Set the hub's address to a DNS name (Settings → Hub) and you never need to do that again.

Operating

High availability

A single hub is a single point of failure: if its VPS stops, every branch and remote user loses the network. With Settings → High availability two or more servers work as one cluster. One node is active and serves every site; the others are standbys that keep a live, exact copy of the hub (a PostgreSQL streaming replica: sites, keys, users, policies, settings, the licence). If the active node stops, a standby takes over by itself and the routers follow.

Building a cluster

  1. Start a cluster on your hub: it becomes node 1 and stays active. Nothing changes for your routers.
  2. Add a node: on the active node press Add a node to get the address and a one-time code. On a fresh SOVA install open Join a cluster, enter them and confirm. The new server copies the cluster's database in a minute and becomes a standby. Its own data is replaced and its administrator accounts become the cluster's.
  3. Repeat for as many nodes as you want. Any node's panel can add, hand over to or remove the others.

When something fails

  • The active node stops answering: after about 30 seconds the first standby promotes itself (with three or more nodes a majority must agree). The administrators are notified.
  • Every router runs a small script. It moves its tunnel only when it is sure the hub died: the hub stopped answering, the site's own Internet works, and another node confirms it is the active one. Then it goes straight to that node, usually within a minute. A site that loses its own line never moves.
  • Port 80 of every node must be reachable from the branches: that is where a router asks which node is active.
  • The old active node comes back: it sees that another node took over, steps back and copies itself from the new active node. Nothing needs doing.
  • Hand over on purpose: Make active on a standby moves the role in seconds with no data loss, for maintenance.
  • Automatic failover off: nothing promotes by itself; Take over now on a standby does it by hand when the active node is gone.

Removing a node, starting over

Remove on any node's page takes a standby out of the cluster. It resets itself to a fresh install and can be added again with a new code.

The administrator accounts and passwords stay on a removed node; its backups are erased.

The active node cannot be removed: make another node active first.

When no other node is left, Leave the cluster turns the hub back into a single hub with all its data, and you can start a new cluster any time.

Licence and updates

One licence covers the whole cluster. It lives in the replicated database and only the active node talks to the licence server, so a cluster needs no extra licences and the licence follows the active node. Update the active node from System → Update; the standbys then update their own program files by themselves (the database is migrated once, on the active node). The page warns while the nodes run different versions.

Good to know

  • A standby serves only the High availability page; sites, users and settings are changed on the active node.
  • Nodes talk to each other over the Internet: PostgreSQL replication (TLS, password, and only from the other nodes' addresses) and a signed HTTP API. Keep the nodes' clocks right (NTP).
  • With three or more nodes, an active node that cannot reach the rest of the cluster stops serving after 20 seconds, so two nodes never serve the sites at the same time.
  • With two nodes, if the link between them breaks while both are alive, the standby may take over. When the link returns the older node steps back and what it saved meanwhile is lost. Use three nodes where that matters.
  • Remote users have the hub's address in their profile: set it to a DNS name that points at the active node (Settings → Hub) and they need nothing after a failover; with an IP address they download their profile again.

Operating

Updates

SOVA checks for a new release every day. System → Update shows the installed version, what is new and the release history; Apply update downloads the release, verifies SecuryTik's signature and checksum, backs up the current version, installs, migrates the database and restarts the services. Tunnels keep running during an update. Unsigned or tampered releases are refused.

Operating

Licensing & plans

Each install is licensed on its own, by the number of sites and remote users. A high-availability cluster is one licence for all its nodes, and moving to a new server moves the licence with it.

PlanSitesRemote usersPrice
Free21free
Lite53$30 / month
Pro106$50 / month
Max2012$100 / month
Unlimitedunlimitedunlimited$175 / month

Yearly billing is 10 × the monthly price (Unlimited: $2,000 a year). System → License: sign in with your SecuryTik account or press Link this device and approve the code on your dashboard. Request a plan from the same page — every requested plan is approved for one month free — then pay online on the dashboard. The licence is signed and verified offline: the hub keeps working through an internet outage, warns before a plan ends and gives a grace period before new sites and users are limited. When the grace period ends without renewal the hub locks and its tunnels stop until the licence is renewed.

Activation is required. A new hub opens only System → License until it is registered with a SecuryTik account; the Free plan starts at registration.

Operating

Cloudflare Tunnel

To reach the panel on your own domain (e.g. vpn.company.com):

  1. In Cloudflare Zero Trust → Networks → Tunnels, create a tunnel and copy its token.
  2. In SOVA, System → Cloudflare Tunnel: paste the token and save — SOVA installs and runs the connector.
  3. In Cloudflare, add a public hostname on the tunnel pointing to http://localhost:80.

Only the panel goes through Cloudflare; the VPN tunnels always connect directly to the hub's public address.

Reference

REST API

System → API → New token (Super Admin). Pick the scopes the token needs — sites, ra, policies, dns, stats, devices, audit, settings, performance, backups, tools:run, hub:read, hub:write, routers:diagnose, routers:write, support:read, services:write, proxy:write, each read or write — an optional expiry and a rate limit. The secret is shown once.

curl -H "Authorization: Bearer sova_…" http://<hub>/api/v1/sites
EndpointPurpose
GET /api/v1/meThe token's scopes and limits.
GET/POST /sites, GET/PATCH/DELETE /sites/{id}Sites with live tunnel state.
PUT/DELETE /sites/{id}/networksAdd, change, remove networks.
GET /sites/{id}/script, POST /sites/{id}/install-linkThe branch script or a 24-hour one-liner.
GET/POST /ra, PATCH/DELETE /ra/{id}, PUT /ra/{id}/accessRemote users and what they can reach.
GET /ra/{id}/wireguard.conf, /ra/{id}/l2tp, /ra/{id}/accountingProfiles and daily online time / traffic.
GET /ra/{id}/wireguard.png or .svgThe WireGuard profile as a QR code, ready to hand to a phone.
GET /ra/{id}/openvpn.ovpn, /ra/{id}/openvpn/ca.crt|client.crt|client.keyOpenVPN profile and the three files IP phones ask for.
POST /proxySend an HTTP(S) request to a device inside a site's network (a PBX, a camera, a printer) and get its reply — for automation that can only use ports 80 and 443. Only addresses inside your sites' networks; redirects are returned, not followed; every call is in the audit log. Needs proxy:write.
GET /hub/licensePlan, end date and days left, for your own reminders.
/policies, /dns, /network-typesPolicies, DNS records, network types.
/stats/overview, /stats/tunnels, /alerts, /audit, /devicesMonitoring data.
GET/PATCH /settingsHub address, DNS zone, DHCP pool and every timing.
/devices/hosts, /forwards, /forwards/hitsDevices, static leases, DNS names, Wake on LAN, port forwards and their hits.
POST /tools/ping|trace|speed, GET /tools/jobs/{job}Tests from the hub or a site, as jobs.
/performance/results, /performance/matrix, /performance/checkResults and Check performance now.
/backupsTake, download, restore and delete backups.
/hub/*Support's view of the hub: every site's tunnels and their history, OSPF neighbours and routes, remote users online, the configuration apply log, config deliveries; ping, traceroute and DNS from the hub; re-apply, sync or retry a delivery.
/routers/{id}/*Read any RouterOS menu and log of a managed branch router, run a test, change a menu or run a script. SOVA's own login and link to the router (sova-mgr, the API service, anything tagged SOVA) stays locked.
/support/*Incidents with their technical details, service logs, host and service facts; restart one hub service (needs services:write).

The interactive reference with every field is at http://<hub>/api/v1/docs. Every write is recorded in the audit log; writes are refused while the licence is locked.

Reference

Webhooks

System → API → Webhooks: add an https URL and the events to send — site.created, site.updated, site.deleted, ra.created, ra.updated, ra.deleted, alert.opened, alert.resolved. Each delivery is a JSON POST signed with the endpoint's secret:

X-Sova-Event: alert.opened
X-Sova-Signature: sha256=<HMAC-SHA256 of the raw body>

Failed deliveries are retried with back-off; an endpoint that keeps failing is disabled. Use the test button to try an endpoint.

Reference

Security model

  • The hub firewall allows only the panel, SSH and the VPN ports; everything between sites passes the hub's policy chain.
  • Private keys and passwords are encrypted at rest with the install's own key (/etc/sova/secret.key); a database dump alone reveals nothing.
  • The panel process has no network privileges; a separate root agent applies validated configuration.
  • Branch scripts contain the branch's keys — install links expire after 24 hours; rotate keys if a script leaks.
  • Updates and licences are signed by SecuryTik and verified offline.

Reference

Troubleshooting

SymptomCheck
Site never connectsOn the router: /interface wireguard peers print — last handshake. Is UDP 51000+N allowed by the VPS provider's firewall? Is the public address in Settings correct?
Script stops with "address conflict"The router already uses the site's block or 172.31.0.0/16. Re-address that LAN or choose another site number.
Tunnel up but no routes/routing ospf neighbor print on the router; Routes page on the hub. Rejected routes mean the branch advertises foreign addresses.
L2TP never comes upUDP 500/4500 blocked, or another L2TP client shares the same public IP.
Remote user connects but reaches nothingGive the user access rules on their page; check policies.
Internal names do not resolveThe client must use 172.31.0.1 as DNS (branch DHCP and profiles do this).
Panel unreachablesystemctl status sova-api nginx; health-watch restarts failed services every minute.

Logs: journalctl -u sova-agent -u sova-monitor -u sova-api. Last configuration apply: System → Settings.

Report this error. An error page's Report this error to support button sends the incident — its reference and technical details, never your data — straight to SecuryTik support. Incident references read SOVA-xxxxxx; quote it when you also write in.

Reference

FAQ

Do branches need a public IP?

No. Branches dial out to the hub; only the hub needs a public IP.

Does branch internet go through the hub?

No. Only company traffic crosses the tunnels. Remote users on Full passthrough are the only ones whose internet goes through the hub.

Can I add a branch later?

Yes — create the site and paste its script. The other branches learn it automatically.

Which routers are supported?

MikroTik RouterOS 7 for branches. Remote users on any device with WireGuard or L2TP/IPsec.

What happens if the hub goes down?

Branches keep their local networks and internet; site-to-site traffic resumes when the hub is back. For a hub that must not go down, run a high-availability cluster: a second server takes over by itself.

Can I move the hub to another server?

Yes, from the panel and without downtime: Move to a new server.

Can SOVA carry our phone system?

Yes. The PBX stays on a private network; branches, desk phones (OpenVPN) and mobile softphones (WireGuard) reach it through the tunnels, and PBXs in different cities trunk to each other inside SOVA. See IP phones & PBX.

Will I be warned before the licence ends?

Yes: alerts 30, 14, 7, 3 and 1 days before the plan ends, on every notification channel you set up.

Need more help?

Email accounts@securytik.com.