What this guide covers
WatchZit is a free, open-source (AGPLv3) monitoring system for networks and servers, with one web console on your own server. There are no licence keys and no price per host or per check; MaxZit sells optional support. This guide does not compare products. It covers the work of a team that watches servers and services, and says, for each part, what WatchZit does and where it stops. Discovery, templates and learned thresholds are in the Zabbix alternative guide; Windows rollout and directory sign-in in the PRTG alternative guide.
- Install and the first host
- Checks you pick from a list
- Servers: agent or SSH and WinRM
- Web applications, certificates and APIs
- Your own triggers
- Notifications, escalation and on-call
- Acknowledge, dependencies and maintenance
- Maintenance and acknowledgements from your scripts
- Dashboards, reports and who may change what
- What WatchZit does not do
- Run WatchZit next to your current monitoring
- What to do next
Install and the first host
Import the appliance from the download page (OVA, qcow2, VHDX or ISO; 2 CPUs, 2 GB of memory, 16 GB disk), or run the install script on a 64-bit Debian 12+, Ubuntu 22.04+, AlmaLinux or Rocky Linux 9+ server:
curl -fsSL https://github.com/Max-Zit/watchzit/releases/latest/download/install.sh | sudo sh
Open https://<server>:8443, enter the setup token, create the first administrator and set up two-factor sign-in. Then add a host under Devices → Add a device: a name, an IP address or host name, its type and group, and Ping on. Every TCP port you enter there becomes a check that connects every minute. Depends on names the router or switch the host is reached through.
That is the whole configuration: there is no file to edit and no reload. Every change is written to the Audit log, with who made it.
Checks you pick from a list
On the device page, Add a check offers these types. Each has its own Interval (seconds) (10 to 3,600; 60 by default) and Timeout (ms); passwords and tokens of checks are stored encrypted and never shown again.
| Check | Measures |
|---|---|
| Ping (ICMP), TCP port | response time and packet loss; whether a port accepts connections, and how fast |
| Web page (HTTP) | status code, response time, a text that must be on the page, days until the certificate expires |
| TLS certificate | days until it expires and whether it is trusted, on any TLS port, with a server name for SNI |
| DNS | a DNS server must answer a name with a record type and, optionally, an expected value |
| PostgreSQL, MySQL / MariaDB | sign-in with a monitoring user, version, connections, response time |
| Web scenario (several steps) | a sign-in and pages behind it, step by step, like a user |
| JSON API value | one number from any JSON API |
| Linux over SSH (no agent), Windows over WinRM (no agent) | CPU, memory, swap or page file, the fullest disk and the uptime |
| Network path (traceroute) | every router on the way with loss and latency, like MTR |
| Hypervisor (vSphere, Proxmox VE) | hosts, virtual machines and datastores |
| SNMP | interfaces, CPU, storage and any numeric OID |
For the same checks on many hosts, a template adds them to all devices of a kind or a group, also to devices added later; under Change selected: in the device list you set the group, the dependency or a template for many hosts at once.
Servers: agent or SSH and WinRM
The WatchZit agent for Windows (MSI) and Linux (deb, rpm; x86-64 and arm64) makes one outbound connection to port 8444, so the servers need no open port. It is enrolled with a one-time code from Agents and updates itself from the console. Every minute it reports CPU, memory, every disk and every interface, and:
- Watch services on the device page: a service that stops opens a problem.
- Watch processes: program names such as
nginx,sqlservr.exeorjava, with how many run, their CPU and memory; none running opens a problem. - Critical and error events of the Windows System and Application logs, and messages of priority err and worse from the systemd journal, in Logs.
- Docker containers on Linux and Hyper-V machines on Windows, every five minutes.
Where nothing may be installed, Linux over SSH (no agent) runs one fixed command that only reads /proc, df and uname, and checks the host key on every run; Windows over WinRM (no agent) signs in with NTLMv2 and encrypted messages, or over HTTPS. Their values have the agent's names, so the same triggers apply. The guide Agentless Linux and Windows server monitoring shows the setup.
Web applications, certificates and APIs
- Web page (HTTP) checks a URL for its status, a text on the page and the days until its certificate expires.
- Web scenario (several steps) goes through an application like a user. One step per line — the method and address, the expected status, a text the page must contain and the form data; cookies are kept from step to step, and
{secret}is the check's password:
It measures the time of all steps and how many passed; the alert names the step that failed.GET /login | 200 | Sign in POST /login | 302 | | user=monitor&password={secret} GET /dashboard | | Welcome - TLS certificate watches mail, LDAP or any other TLS port. The built-in triggers Certificate expires soon (fewer than 21 days) and Certificate expires (fewer than 7) apply to it.
- JSON API value reads one number from a health or status endpoint of your own services: an address, a path such as
results.cpu.0.current, and optional Bearer or basic sign-in.
Your own triggers
Built-in triggers already watch devices that are down, checks that are down, high CPU load, full memory and disks, packet loss and expiring certificates. Your own go under Problems → Triggers:
- Value out of range: a value (CPU load, response time, connections …), above or below a threshold, for at least some minutes, so a short dip does not open a problem; with a Severity and Applies to all devices, a group or one device.
- One check: the bell button next to a check on the device page adds a trigger for that check alone, for example one important disk or the response time of one web page.
- No new data: a check that got no new value for the minutes you set — an agent that stopped reporting.
- Syslog message or trap: a text or a regular expression in syslog messages, traps or the agents' system logs.
A problem closes by itself when the condition ends. Triggers that learn what is normal and forecasts are explained in the Zabbix alternative guide.
Notifications, escalation and on-call
Under Settings → Notifications, administrators add channels: email over your SMTP server, a JSON webhook (signed with HMAC-SHA256 in X-WatchZit-Signature when you set a key), Slack, Microsoft Teams, Discord, Telegram, ntfy, PagerDuty (problems open and resolve incidents), and SMS through Twilio or your own gateway. Each channel has a lowest severity and, with Also when resolved, also reports when problems are resolved.
Under Escalation and hours of duty of a channel:
- Tell after (minutes): the channel hears only of problems still open after that time — the team lead after 15 minutes when the first line has not reacted.
- Repeat every (minutes): tells it again while the problem stays open. Acknowledged problems are not escalated or repeated, unless Also problems someone acknowledged is on.
- Only on some days and hours (on-call): for example SMS to the night shift Monday to Friday from 18 to 08 and all weekend. A problem that starts outside those hours reaches the channel when its duty starts, if it is still open.
- A rotation is one channel per person or team, each with its own days and hours.
A channel that was told about a problem is also told when it is resolved; the others are not.
Acknowledge, dependencies and maintenance
- Problems lists what is open, worst first. Acknowledge a problem with a note so the others know someone is on it. Resolved problems stay in the history for a year.
- Dependencies: with Depends on set, a host behind a router that is down is only unreachable; its problems are recorded but not sent.
- Maintenance under Problems → Maintenance for devices or groups: they are still measured, but nothing is sent; problems still open when the window ends are sent then. Repeat brings the window back every day, week or month, optionally until a last day. Availability reports show maintenance time apart.
Maintenance and acknowledgements from your scripts
Create a token under My account → API tokens; it acts with your role and is shown once. A deploy script can then plan a maintenance window before it restarts the servers:
curl -H "Authorization: Bearer wzat_…" -H "Content-Type: application/json" \
-d '{"name":"Deploy","starts_at":"2026-11-20T22:00:00+01:00","ends_at":"2026-11-20T22:30:00+01:00","group_ids":[3]}' \
https://monitor.example.com:8443/api/v1/maintenance
The API also reads the status, devices, checks, graphs and problems, and acknowledges problems with POST /api/v1/problems/{id}/ack. Changes made with a token are written to the audit log with the token's name.
Dashboards, reports and who may change what
- Dashboards with graphs, current values, device states, open problems, maps and graphs from data sources; they reload every minute for a wall screen.
- Reports: availability per device for 7, 30 or 90 days or a month, with downtime and the number of problems, as CSV or PDF, with an SLA target (%) per group and by email every week or month.
- Roles: Administrator, Operator (devices, checks, maps, triggers, maintenance, dashboards) and Viewer. Sign-in with two-factor, Active Directory / LDAP or single sign-on (OpenID Connect, SAML 2.0).
What WatchZit does not do
- No configuration files and no importer. Hosts, checks and triggers are set up in the console; existing configurations cannot be read in, and history starts on the day you add a host.
- No check that runs your own scripts or plugins. Values of your own come from a JSON API value check against a service of yours, an SNMP OID, or a data source on a graph.
- No results pushed from outside. The API reads data, acknowledges problems and plans maintenance; it does not create hosts, checks or triggers.
- The agentless checks cover the basics (CPU, memory, swap or page file, the fullest disk, uptime). Every drive, services, processes and logs need the agent.
- The database checks are PostgreSQL and MySQL/MariaDB; a web scenario has 1 to 10 steps.
- The server runs on Linux (x86-64) with PostgreSQL, or as the appliance.
Run WatchZit next to your current monitoring
- Install WatchZit on its own VM; nothing on the existing monitoring server changes.
- Start with one group of hosts. Use the agentless checks or install the WatchZit agent next to what runs on the servers today; the agent needs only its outbound connection to port 8444.
- Give WatchZit its own monitoring users (SSH, WinRM, databases, web scenarios), so you can tell its sign-ins apart and remove them later.
- Send WatchZit's problems to a test channel first, not to the on-call phones.
- Compare for a few weeks which problems each system raised and which ones mattered, then decide which one pages the on-call team.
What to do next
- See everything WatchZit does on the features page and get it on the download page.
- Network devices over SNMP: A LibreNMS and Observium alternative.
- Branch offices behind NAT: Monitor branch offices with a remote collector.
- The complete reference is the user guide built into the console; see the documentation.
Questions and ideas are welcome in GitHub Discussions; companies can get commercial support from MaxZit.
Nagios and Checkmk are trademarks of their respective owners. MaxZit is not affiliated with, sponsored or endorsed by the owners of Nagios or Checkmk; the names are used only to describe what this guide is about.