You are viewing documentation for the development version, it may be incomplete.
Join our translation project and help translate Zabbix documentation into your native language.

1 Configuring a network discovery rule


To configure a network discovery rule used by Zabbix to discover hosts and services:

  • Go to Data collection → Discovery
  • Click on Create discovery rule (or on the rule name to edit an existing one)
  • Edit the discovery rule attributes

Rule attributes

All mandatory input fields are marked with a red asterisk.

Parameter Description
Name Unique name of the rule. For example, "Local network".
Discovery by Discovery is performed by:
Server - by Zabbix server
Proxy - by Zabbix proxy (selected in the proxy name field)
IP range The range of IP addresses for discovery. It may have the following formats:
Single IP:
Range of IP addresses: 192.168.1-10.1-255. The range is limited by the total number of covered addresses (less than 64K).
IP mask:
supported IP masks:
/16 - /30 for IPv4 addresses
/112 - /128 for IPv6 addresses
This field supports spaces, tabulation and multiple lines.
Update interval This parameter defines how often Zabbix will execute the rule.
The interval is measured after the execution of previous discovery instance ends so there is no overlap.
Time suffixes are supported, e.g. 30s, 1m, 2h, 1d.
User macros are supported.
Note that if a user macro is used and its value is changed (e.g. 1w → 1h), the next check will be executed according to the previous value (far in the future with the example values).
Maximum concurrent checks per type Set the maximum number of discovery threads (workers) per service check for parallel processing of discovery checks:
One - one thread
Unlimited - unlimited number of threads (but no more than in the StartDiscoverers parameter)
Custom - set a custom number of threads (0-999)
Note that all discovery rules with SNMPv3 asynchronous service checks are processed by one worker due to the peculiarities of libsnmp implementation, i.e. increasing the number of workers will not increase the discovery speed.
Checks Zabbix will use this list of checks for discovery. Click on to configure a new check in a popup window.
Supported checks: SSH, LDAP, SMTP, FTP, HTTP, HTTPS, POP, NNTP, IMAP, TCP, Telnet, Zabbix agent, SNMPv1 agent, SNMPv2 agent, SNMPv3 agent, ICMP ping.
A protocol-based discovery uses the net.tcp.service[] functionality to test each host, except for SNMP which queries an SNMP OID. Zabbix agent is tested by querying an item in unencrypted mode. Please see agent items for more details.
The 'Ports' parameter may be one of following:
Single port: 22
Range of ports: 22-45
List: 22-45,55,60-70
Since Zabbix 7.0. all service checks are performed asynchronously, except for LDAP checks.
Since Zabbix 7.0 HTTP/HTTPs checking is done via libcurl. If Zabbix server/proxy is compiled without libcurl, then HTTP checks will work like in previous versions (i.e. as TCP checks), but HTTPS checks will not work.
Device uniqueness criteria Uniqueness criteria may be:
IP address - no processing of multiple single-IP devices. If a device with the same IP already exists it will be considered already discovered and a new host will not be added.
<discovery check> - either Zabbix agent or SNMP agent check.
Host name Set the technical host name of a created host using:
DNS name - DNS name (default)
IP address - IP address
<discovery check> - received string value of the discovery check (e.g. Zabbix agent, SNMP agent check)
See also: Host naming.
Visible name Set the visible host name of a created host using:
Host name - technical host name (default)
DNS name - DNS name
IP address - IP address
<discovery check> - received string value of the discovery check (e.g. Zabbix agent, SNMP agent check)
See also: Host naming.
Enabled With the check-box marked the rule is active and will be executed by Zabbix server.
If unmarked, the rule is not active. It won't be executed.
Exceeding file descriptor limit

In case of large number of concurrent checks it is possible to exhaust the file descriptor limit for the discovery manager.

The number of file descriptors required for detection equates to the number of discovery workers * 1000. By default there are 5 discovery workers, while the soft limit of the system is ~1024.

In this case Zabbix will reduce the default number of concurrent checks per type for each worker and write a warning to the log file. But, if the user has set a higher value for Maximum concurrent checks per type than the value calculated by Zabbix, Zabbix will use the user value for one worker.

A real life scenario

In this example, we would like to set up network discovery for the local network having an IP range of

In our scenario we want to:

  • discover those hosts that have Zabbix agent running
  • run discovery every 10 minutes
  • add a host to monitoring if the host uptime is more than 1 hour
  • remove hosts if the host downtime is more than 24 hours
  • add Linux hosts to the "Linux servers" group
  • add Windows hosts to the "Windows servers" group
  • use the template Linux for Linux hosts
  • use the template Windows for Windows hosts
Step 1

Defining a network discovery rule for our IP range.

Zabbix will try to discover hosts in the IP range of by connecting to Zabbix agents and getting the value from the system.uname key. The value received from the agent can be used to name the hosts and also to apply different actions for different operating systems. For example, link Windows servers to the template Windows, Linux servers to the template Linux.

The rule will be executed every 10 minutes.

When this rule is added, Zabbix will automatically start the discovery and generation of the discovery-based events for further processing.

Step 2

Defining a discovery action for adding the discovered Linux servers to the respective group/template.

The action will be activated if:

  • the "Zabbix agent" service is "up"
  • the value of system.uname (the Zabbix agent key we used in rule definition) contains "Linux"
  • Uptime is 1 hour (3600 seconds) or more

The action will execute the following operations:

  • add the discovered host to the "Linux servers" group (and also add host if it wasn't added previously)
  • link host to the Linux template. Zabbix will automatically start monitoring the host using items and triggers from the "Linux" template.
Step 3

Defining a discovery action for adding the discovered Windows servers to the respective group/template.

Step 4

Defining a discovery action for removing lost servers.

A server will be removed if "Zabbix agent" service is 'down' for more than 24 hours (86400 seconds).