Zabbix vs Checkmate

These two sit at opposite ends of the monitoring catalogue and share only a licence. Zabbix is enterprise infrastructure monitoring: agents, SNMP, IPMI, JMX, templates, network auto-discovery, escalation chains, proxies for remote sites. Checkmate is a modern uptime monitor with an unusually good dashboard that you bring up with a compose file. Both are AGPL-3.0 and free to self-host with no user cap, and that is roughly where the resemblance ends. The real question is whether you are monitoring services or monitoring an estate.

What Zabbix actually asks of you

Zabbix is not a container you start and forget. A working install is four pieces:

• the Zabbix server (C), which does the collecting and the alerting
• a database — MySQL, MariaDB, Percona or PostgreSQL, chosen at install time, because the package name encodes it
• a PHP frontend behind a web server
• an agent on every host you want to watch

Upstream ships packages from its own apt repository, so the install is a repo .deb, an apt install, a database and user created by hand, a schema loaded from a gzipped SQL script, a password set in zabbix_server.conf, and the nginx listen and server_name lines uncommented before anything starts. Official container images exist, but they keep the same split across server, frontend, database and agent images — you trade package management for orchestration rather than escaping the complexity. It rates 4 out of 5 to deploy, and the catalogue estimates 4 GB of RAM, most of it going to the database.

What Checkmate asks of you

Checkmate is two moving parts: the application image and a MongoDB service. The reference compose file starts both. You need a JWT_SECRET in the environment — openssl rand -hex 32 is enough — and, if you will reach it at a domain or a LAN address rather than localhost, CLIENT_HOST set to that origin. It serves on port 52345 and terminates no TLS itself, so a reverse proxy goes in front of anything public. That is the whole install: 2 out of 5, an estimated 2 GB of RAM, and you are looking at a dashboard the same afternoon.

Be clear that MongoDB is not embedded in the app image. It is a second service to run and back up, which is one more than Uptime Kuma needs and three fewer than Zabbix.

Coverage is the reason Zabbix exists

If the two covered the same ground, none of Zabbix's weight would be defensible. They do not.

Checkmate monitors HTTP endpoints, ping, ports, SSL certificates, Docker containers and game servers, and adds PageSpeed reports and status pages. Hardware metrics are not built in — those come from Capture, a separate agent you install on each machine you want to watch.

Zabbix monitors things Checkmate cannot see at all: switches, routers, UPSes and printers over SNMP, server hardware out-of-band over IPMI, Java applications over JMX, plus agentless checks and its own agents. On top of that sit the features that only make sense at scale — templates so a new host inherits a hundred items at once, network auto-discovery so hosts add themselves, escalation chains so an unacknowledged alert climbs to the next person, and proxies so a remote site keeps collecting when the link to head office drops.

That is the trade, stated plainly: Checkmate covers what you publish, Zabbix covers what you own.

Which should you self-host?

Pick Checkmate if…

• You are watching services and sites, and the checks you need are HTTP, ping, port, SSL and Docker.
• You want a good-looking dashboard and status pages without designing anything.
• Two containers is your ceiling for a monitoring stack, and PageSpeed reports are a welcome bonus.

Pick Zabbix if…

• Your estate includes network gear, hardware, or JVMs — SNMP, IPMI and JMX are not optional for you.
• You have enough hosts that templates and auto-discovery pay for themselves, rather than being ceremony.
• You need escalation chains, remote-site proxies and long-term trend storage, and you have someone whose job includes owning this.

Running either on a VPS

The sizing gap is the practical one, and it is large. Checkmate's app plus MongoDB is comfortable on a mid-tier instance and leaves room for other things. Zabbix wants its own box: the estimate here is 4 GB, and upstream's smallest published sizing example is 2 CPU cores and 8 GiB for roughly a thousand monitored metrics — so size it against your metric count, not against an uptime checker. Neither figure is a hard floor published by the projects; both are starting points. Put TLS in front of whichever you choose, and if you are watching a handful of servers rather than an estate, look hard at whether you want Zabbix at all before you start creating databases. Setup steps and VPS options are linked below.