Monitoring TiDB with Zabbix Server 5.4 – Step‑by‑Step Guide

The Zabbix official integration page lists a TiDB monitoring solution that can be applied with Zabbix Server 5.4. The monitoring logic works as follows:

Zabbix Server calls TiDB’s metrics endpoint (e.g., http://127.0.0.1:10080/metrics) via an HTTP agent.

The built‑in Prometheus to JSON feature (available from version 4.2) converts the raw Prometheus data into JSON format.

Using JSONPath, any desired metric—such as the total response time of all queries—can be extracted.

All of these steps are packaged in ready‑made templates, but they require Zabbix ≥ 5.4 and a compatible operating system (Server on RHEL 8/CentOS 8, agents on CentOS 6‑8). Older Zabbix versions have limited compatibility, so for other versions you should use at least 4.2 and manually recreate items, triggers, etc., based on the 5.4 defaults.

Installation

Download and install Zabbix 5.4 following the official documentation (e.g., Zabbix 5.4 download page ).

Adding a Host

Navigate to Configuration → Host and create a new host. Remember to manually add the HTTP‑agent interface shown in the red box of the original UI.

Linking the TiDB Template

Associate the host with the three provided TiDB templates: TiDB by HTTP, TiDB PD by HTTP, and TiDB TiKV by HTTP. Only the components installed on the host need to be linked.

Modifying Macros

Two macros in the template must be customized: {$TIKV.PORT} – set to the actual TiKV port (if the default port is used, no change is needed). {$TIKV.URL} – by default set to localhost; replace it with the real address, e.g., 10.186.64.59:20180/metrics. You can use the built‑in macro {HOST.IP} to automatically insert the host’s IP.

Both macros are defined in the template’s macro section; {$TIKV.URL} must be a concrete value, while {HOST.IP} is resolved automatically.

Viewing Collected Data

Open the Monitoring → Latest data page; if metrics appear for the host, the configuration is successful.

Keywords: TiDB, Zabbix, monitoring.