Install Zabbix agent 2 on Linux: passive and active checks

You need both VMs running: zbx01 (192.168.56.80) and web01 (192.168.56.81). Each step says where to run the command. The frontend is at https://zbx01.lab, user Admin, password Zbx@Secure2026.

1Install agent 2 on web01

Run this on web01:

cd /tmp
wget https://repo.zabbix.com/zabbix/8.0/release/ubuntu/pool/main/z/zabbix-release/zabbix-release_latest_8.0+ubuntu26.04_all.deb
sudo dpkg -i zabbix-release_latest_8.0+ubuntu26.04_all.deb
sudo apt update
sudo apt install -y zabbix-agent2
zabbix_agent2 -V | head -1

You should see:

zabbix_agent2 (Zabbix) 8.0.0

Use the agent from the Zabbix repository, not Ubuntu’s own zabbix-agent2 package, so the agent version matches the server. An older agent works with a newer server, but you miss new item keys.

✅ Checkpoint: zabbix_agent2 -V reports version 8.0.0 on web01.

2Set Server, ServerActive and Hostname

Keep the shipped file as it is and put your settings in a small file in the include directory. It’s easier to read, survives package upgrades cleanly and is how config management tools usually do it. The later value wins when a parameter appears twice, so first comment out the three shipped lines.

Run this on web01:

sudo sed -i -E 's/^(Server|ServerActive|Hostname)=/# &/' /etc/zabbix/zabbix_agent2.conf
sudo tee /etc/zabbix/zabbix_agent2.d/10-lab.conf >/dev/null <<'EOF'
Server=192.168.56.80
ServerActive=192.168.56.80
Hostname=web01
EOF
sudo grep -E '^(Server|ServerActive|Hostname)=' /etc/zabbix/zabbix_agent2.conf /etc/zabbix/zabbix_agent2.d/10-lab.conf

You should see only the lines from 10-lab.conf, because the shipped ones now start with #:

/etc/zabbix/zabbix_agent2.d/10-lab.conf:Server=192.168.56.80
/etc/zabbix/zabbix_agent2.d/10-lab.conf:ServerActive=192.168.56.80
/etc/zabbix/zabbix_agent2.d/10-lab.conf:Hostname=web01

The shipped file had Server=127.0.0.1, ServerActive=127.0.0.1 and Hostname=Zabbix server, which is right only on the Zabbix server itself. The Include line reads every .conf file in zabbix_agent2.d, and plugin settings live one level down in plugins.d.

✅ Checkpoint: Only 10-lab.conf sets Server, ServerActive and Hostname, and the shipped lines are commented out.

3Test the config and start the agent

Run this on web01:

sudo zabbix_agent2 -T -c /etc/zabbix/zabbix_agent2.conf && echo CONFIG OK
sudo systemctl restart zabbix-agent2
sudo systemctl enable zabbix-agent2
sudo ss -ltnp | grep 10050
sudo grep -E 'Starting Zabbix Agent 2|hostname' /var/log/zabbix/zabbix_agent2.log | tail -2

The test, ss and grep commands print this (your PID and times will differ):

Validating configuration file "/etc/zabbix/zabbix_agent2.conf"
Validation successful
CONFIG OK
LISTEN 0      4096               *:10050            *:*    users:(("zabbix_agent2",pid=2291,fd=8))
2026/11/08 11:42:10.118342 Starting Zabbix Agent 2 (8.0.0)
2026/11/08 11:42:10.119015 Zabbix Agent2 hostname: [web01]

-T validates the config without starting the service, so a typo shows up here and not as a silent restart loop. The log confirms the hostname the agent will use for active checks. Agent 2 log lines start with the date and time, unlike the server’s PID-first format.

✅ Checkpoint: zabbix-agent2 is running, listening on port 10050, and its log shows hostname [web01].

4Open port 10050 on web01 for the server only

Run this on web01:

sudo ufw allow from 192.168.56.0/24 to any port 22 proto tcp
sudo ufw allow from 192.168.56.80 to any port 10050 proto tcp comment 'Zabbix server passive checks'
sudo ufw --force enable
sudo ufw status | grep -E '22|10050'

You should see:

Rules updated
Rules updated
Firewall is active and enabled on system startup
22/tcp                     ALLOW       192.168.56.0/24
10050/tcp                  ALLOW       192.168.56.80              # Zabbix server passive checks

Only zbx01 may reach the agent port. The agent’s Server parameter is a second layer: even if the firewall were open, the agent would refuse anyone else.

✅ Checkpoint: ufw on web01 allows 10050 only from 192.168.56.80.

5Test from the server with zabbix_get

Run this on zbx01:

sudo apt install -y zabbix-get
zabbix_get -s 192.168.56.81 -k agent.ping
zabbix_get -s 192.168.56.81 -k agent.version
zabbix_get -s 192.168.56.81 -k 'vfs.fs.size[/,pused]'
zabbix_get -s 192.168.56.81 -k 'system.cpu.load[all,avg1]'

You should see this (the last two values are live and will differ):

1
8.0.0
48.534803
0.397949

agent.ping always returns 1 when the agent answers. The other values are live: about 48% of web01’s root file system is used, and the one-minute load average is 0.4. If any of these fail, fix it here before you touch the frontend.

✅ Checkpoint: zabbix_get from zbx01 returns 1 for agent.ping and 8.0.0 for agent.version.

6Create the host in the frontend

Open Data collection > Hosts and click Create host. Fill in:

Click Add. The host appears in the list with a grey ZBX icon, which turns green within a minute or two once the first passive check succeeds.

Data collection > Hosts: web01 with the Linux by Zabbix agent template and a green ZBX icon

✅ Checkpoint: web01 is listed in Data collection > Hosts with a green ZBX icon.

7Look at Latest data

Open Monitoring > Latest data, set the filter to host web01 and click Apply. Within a few minutes you’ll see dozens of items with values and small graphs.

Some items in the Linux template are discovered (file systems, network interfaces, block devices), so they appear a little later than the rest, after the discovery rule’s first run. Click the graph link at the end of a row to see the item’s history.

✅ Checkpoint: Latest data for web01 shows CPU utilization, load average and FS [/] space used with recent values.

8Watch the server poll the agent

Run this on zbx01:

sudo grep web01 /var/log/zabbix/zabbix_server.log | tail -3

You should see this line (your PID and time will differ), or nothing, which is also fine:

  4630:20261108:114402.551 enabling Zabbix agent checks on host "web01": interface became available

The server only logs availability changes, not every successful check, so a quiet log is a healthy log. This line appears when the interface first works or comes back after a failure.

✅ Checkpoint: You know where to find availability changes for web01 in the server log.

9Switch web01 to active checks

Open web01 in Data collection > Hosts. On the Templates line click Unlink next to Linux by Zabbix agent (not Unlink and clear, so the history stays), add Linux by Zabbix agent active, and click Update. Keep the agent interface; Zabbix uses it for the host’s IP in macros and for any passive items you add later.

Then check on web01 that the agent picked up its list:

sleep 30
sudo grep -iE 'active check|started to fail|working again' /var/log/zabbix/zabbix_agent2.log | tail -3

You should see no “started to fail” lines; if there were earlier failures you’ll see:

2026/11/08 11:58:31.204871 [101] active check configuration update from [192.168.56.80:10051] is working again

Hover over the ZBX icon: the active check status shows available. In Latest data the same item names now show “Zabbix agent (active)” as the type. Agent 2 asks for a fresh list every 5 seconds by default (RefreshActiveChecks=5), and since 6.4 it only downloads the full list when the configuration changed, so this is cheap.

Older guides mention 120 seconds, which was the default before 6.4.

✅ Checkpoint: web01 uses Linux by Zabbix agent active, and Latest data keeps updating.