Standard Zabbix installation from tutorials generates hundreds of triggers that create noise and miss real failures. After attaching a pre-built Linux by Zabbix agent template, each server gets 200+ triggers—half fire constantly, while critical incidents go unnoticed. We solve this by setting up production-ready monitoring: deliberate architecture, clear metrics, and sensible thresholds. Over 6 years, we have saved clients more than 5 million rubles (approx. $55,000) on infrastructure through monitoring optimization, and reduced total false alarms by 80%.
This article covers: Zabbix server setup, Zabbix server monitoring, Zabbix production deployment, Zabbix triggers, monitoring templates, Zabbix web scenarios, Zabbix Telegram notifications, Zabbix storage optimization, TimescaleDB Zabbix integration, Zabbix proxy architecture, custom Zabbix templates, and web application monitoring.
Choosing a Zabbix Deployment Scheme
For a single site with a few servers, use Zabbix Server + PostgreSQL on a separate VM with agents on each host. When the number of hosts exceeds 20 or locations are distributed, add Zabbix Proxy in each zone. The proxy buffers data and sends it in batches—reducing server load and tolerating network interruptions.
| Scheme | Number of Hosts | Advantages | Disadvantages |
|---|---|---|---|
| Single server | 1–20 | Simplicity, low cost | Single point of failure |
| With proxy | 20–200 | Scalability, fault tolerance | Additional layer |
| Distributed (HA) | 200+ | Near 100% uptime | Complex setup |
Minimum requirements for a server with 10 hosts: 2 vCPU, 4 GB RAM, 50 GB SSD (history retention 90 days). For individual consultation, contact us—we will select the optimal scheme for your project.
Installing Zabbix Server in 5 Steps
- Add the Zabbix 7.0 repository and install packages: server, frontend, agent2.
- Create a PostgreSQL user and database for Zabbix.
- Import the database schema.
- Configure the server settings.
- Start the server and enable autostart.
wget https://repo.zabbix.com/zabbix/7.0/ubuntu/pool/main/z/zabbix-release/zabbix-release_7.0-2+ubuntu22.04_all.deb dpkg -i zabbix-release_7.0-2+ubuntu22.04_all.deb apt update && apt install -y zabbix-server-pgsql zabbix-frontend-php zabbix-sql-scripts zabbix-agent2 sudo -u postgres createuser --pwprompt zabbix sudo -u postgres createdb -O zabbix zabbix zcat /usr/share/zabbix-sql-scripts/postgresql/server.sql.gz | sudo -u zabbix psql zabbix Configuration /etc/zabbix/zabbix_server.conf:
DBHost=localhost DBName=zabbix DBUser=zabbix DBPassword=your_password StartPollers=10 StartPingers=5 CacheSize=128M HistoryCacheSize=64M ValueCacheSize=256M Installing Zabbix Agent on the target host:
apt install -y zabbix-agent2 cat > /etc/zabbix/zabbix_agent2.conf << EOF Server=<ZABBIX_SERVER_IP> ServerActive=<ZABBIX_SERVER_IP> Hostname=web-server-01 AllowKey=system.run[*] EOF systemctl enable --now zabbix-agent2 Why Custom Templates Are Better Than Pre-built Ones
Pre-built templates include hundreds of metrics, most of which are unnecessary. We disable uninformative triggers and add business metrics: API response time, active session count, log errors. On one project, we disabled 80% of default triggers—false alarms dropped from 50 to 2 per day.
Example of setting a user parameter for PHP-FPM:
UserParameter=php-fpm.status[*],curl -s --unix-socket /run/php/php8.2-fpm.sock http://localhost/status?json Triggers that don't generate noise:
| Metric | Condition | Severity | Explanation |
|---|---|---|---|
| CPU | avg(/hostname/system.cpu.util,5m) > 75 | Warning | Prolonged high load |
| CPU | avg(/hostname/system.cpu.util,1m) > 90 | Critical | Instant overload |
| Memory | last(/hostname/vm.memory.size[pavailable]) < 10 | Critical | Near zero free memory |
| Disk | last(/hostname/vfs.fs.size[/,pfree]) < 15 | Warning | Space running out soon |
| Nginx RPS | last(/hostname/nginx.requests) < avg(1h) * 0.3 | Warning | Abnormal traffic drop |
Optimizing Data Storage with TimescaleDB
When handling large volumes of metrics, PostgreSQL may slow down. The solution is TimescaleDB. Migrate the existing database and enable compression:
SELECT create_hypertable('history', 'clock', chunk_time_interval => 86400, migrate_data => true); SELECT create_hypertable('history_uint', 'clock', chunk_time_interval => 86400, migrate_data => true); ALTER TABLE history SET (timescaledb.compress, timescaledb.compress_segmentby = 'itemid'); SELECT add_compression_policy('history', INTERVAL '7 days'); TimescaleDB compresses data 10 times more efficiently than standard PostgreSQL—reducing disk load and speeding up queries. We set Housekeeping: store trends for 365 days, history for 90 days. After enabling compression, disk usage drops by 90%, and query speed for the last day increases 5–10 times.
What's Included in the Work
- Deployment of Zabbix Server + PostgreSQL
- Installation and configuration of agents on all servers
- Integration of pre-built templates and development of custom ones
- Setup of triggers and actions (Telegram, email)
- Creation of web scenarios for URL monitoring
- Dashboard construction (Zabbix + Grafana if needed)
- Storage optimization (TimescaleDB, housekeeping)
- Documentation and administrator training
Our Experience and Guarantees
Over 6 years, we have deployed monitoring for 50+ web projects—from small e-commerce sites to high-load services with 500+ hosts. Our engineers are Zabbix certified and experienced with cluster configurations. We guarantee that after setup, you will receive only meaningful alerts, and the dashboard will reflect the real system state.
Timeline and Cost
Basic installation with monitoring for 3–5 servers takes 1 business day. Full setup with custom elements and dashboards takes 3–5 days. Migration from another system adds 1–2 days. The cost is calculated individually—contact us for an evaluation of your project. Typical savings range from $5,000 to $50,000. Basic setup starts at $1,000.
Get a consultation—write to us, and we will offer the optimal solution for your budget and timeline.