Installation from sources

You can get the very latest version of Zabbix by compiling it from the sources. See also Getting Zabbix source code.

A step-by-step tutorial for installing Zabbix from the sources is provided here.

1 Installing Zabbix daemons

1 Download the source archive

Go to the Zabbix download page and download the source archive. Once downloaded, extract the sources, by running:

tar -zxvf zabbix-8.0.0.tar.gz

2 Create user account

All Zabbix daemon processes run under unprivileged system users. If a Zabbix daemon is started from an unprivileged user account, it will continue running as that user.

In the default configuration, if a daemon is started as root, it will switch to the zabbix user account, which must be present. To create a zabbix system user and group, run the commands listed below.

RedHat-based system:

groupadd --system zabbix
       useradd --system -g zabbix -d /usr/lib/zabbix -s /sbin/nologin -c "Zabbix Monitoring System" zabbix

Debian-based system:

addgroup --system --quiet zabbix
       adduser --quiet --system --disabled-login --ingroup zabbix --home /var/lib/zabbix --no-create-home zabbix

There is no need to create a separate user account for Zabbix frontend.

Security recommendation

If the Zabbix server and agent run on the same machine, it is recommended to run them under separate user accounts. Running both as the same user allows the agent to access the server's configuration file, which could expose sensitive information—such as the database password—to any Admin-level user in Zabbix.

Home directory (optional)

Zabbix processes do not require a home directory, so creating one is generally not recommended. However, if you need functionality that requires a home directory (for example, storing MySQL credentials in $HOME/.my.cnf), you can create it using the commands listed below.

On RedHat-based systems, run:

mkdir -m u=rwx,g=rwx,o= -p /usr/lib/zabbix
       chown zabbix:zabbix /usr/lib/zabbix

On Debian-based systems, run:

mkdir -m u=rwx,g=rwx,o= -p /var/lib/zabbix
       chown zabbix:zabbix /var/lib/zabbix

3 Create Zabbix database

For Zabbix server and proxy daemons, as well as Zabbix frontend, a database is required. It is not needed to run Zabbix agent.

SQL scripts are provided for creating database schema and inserting the dataset. Zabbix proxy database needs only the schema while Zabbix server database requires also the dataset on top of the schema.

Having created a Zabbix database, proceed to the following steps of compiling Zabbix.

4 Configure the sources

C99 with GNU extensions is required for building Zabbix server, Zabbix proxy or Zabbix agent. This version can be explicitly specified by setting CFLAGS="-std=gnu99":

export CFLAGS="-std=gnu99"

When configuring the sources for a Zabbix server or proxy, you must specify the database type to be used. Only one database type can be compiled with a server or proxy process at a time.

To see all of the supported configuration options, inside the extracted Zabbix source directory run:

./configure --help

To configure the sources for a Zabbix server and agent, you may run something like:

./configure --enable-server --enable-agent --with-mysql --enable-ipv6 --with-net-snmp --with-libcurl --with-libxml2 --with-openipmi --with-ares

To configure the sources for a Zabbix server (with PostgreSQL etc.), you may run:

./configure --enable-server --with-postgresql --with-net-snmp

To configure the sources for a Zabbix proxy (with SQLite etc.), you may run:

./configure --prefix=/usr --enable-proxy --with-net-snmp --with-sqlite3 --with-ssh2

To configure the sources for a Zabbix agent, you may run:

./configure --enable-agent

or, for Zabbix agent 2:

./configure --enable-agent2

Notes on compilation options:

• --enable-agent - compiles Zabbix agent, as well as Zabbix get and Zabbix sender command-line utilities.
• --with-libcurl - required for virtual machine monitoring, SMTP authentication, and web.page.* Zabbix agent items. See also: Requirements (libcurl).
• --with-libxml2 - required for virtual machine monitoring.
• --with-libpcre2[=DIR] - Zabbix always compiles with the PCRE2 library; this option only allows to specify a custom PCRE2 installation path.
• --with-mysql=/path/to/mysql_config - specifies the path to a particular MySQL client library configuration. Useful when multiple versions of MySQL or MariaDB are installed.
• --enable-static - statically links libraries (not supported on Solaris). Use this if you plan to distribute compiled binaries to systems without the required libraries. Not recommended when building Zabbix server. To build the server statically, a static version of every external library is required. The configure script does not check for this automatically.
• --with-stacksize=<value> - sets the per-thread stack size in kilobytes (e.g., --with-stacksize=512). You can increase this value if Zabbix crashes or freezes due to stack overflows (e.g., during preprocessing on systems with low default thread stack limits).

checking for main in -lmysqlclient... no
       configure: error: Not found mysqlclient library

/usr/bin/ld: cannot find -lssl
       /usr/bin/ld: cannot find -lcrypto

See also:

• Compiling Zabbix with encryption support
• Known compilation issues

5 Make and install everything

make install

This step should be run as a user with sufficient permissions (commonly 'root', or by using sudo).

Running make install will by default install the daemon binaries (zabbix_server, zabbix_agentd, zabbix_proxy) in /usr/local/sbin and the client binaries (zabbix_get, zabbix_sender) in /usr/local/bin.

6 Review and edit configuration files

• edit the Zabbix agent configuration file /usr/local/etc/zabbix_agentd.conf

You need to configure this file for every host with zabbix_agentd installed.

You must specify the Zabbix server IP address in the file. Connections from other hosts will be denied.

• edit the Zabbix server configuration file /usr/local/etc/zabbix_server.conf

You must specify the database name, user and password (if using any).

The rest of the parameters will suit you with their defaults if you have a small installation (up to ten monitored hosts). You should change the default parameters if you want to maximize the performance of Zabbix server (or proxy) though.

• if you have installed a Zabbix proxy, edit the proxy configuration file /usr/local/etc/zabbix_proxy.conf

You must specify the server IP address and proxy hostname (must be known to the server), as well as the database name, user and password (if using any).

7 Start up the daemons

Run zabbix_server on the server side.

zabbix_server

Run zabbix_agentd on all the monitored machines.

zabbix_agentd

If you have installed Zabbix proxy, run zabbix_proxy.

zabbix_proxy

2 Installing Zabbix web interface

Copying PHP files

Zabbix frontend is written in PHP, so to run it a PHP supported webserver is needed. Installation is done by simply copying the PHP files from the ui directory to the webserver HTML documents directory.

Common locations of HTML documents directories for Apache web servers include:

• /usr/local/apache2/htdocs (default directory when installing Apache from source)
• /srv/www/htdocs (OpenSUSE, SLES)
• /var/www/html (Debian, Ubuntu, Fedora, RHEL)

It is recommended to use a subdirectory instead of the HTML root. To create a subdirectory and copy Zabbix frontend files into it, execute the following commands, replacing the <htdocs> with actual directory:

mkdir <htdocs>/zabbix
       cd ui
       cp -a . <htdocs>/zabbix

If planning to use any other language than English, see Installation of additional frontend languages for instructions.

Installing frontend

Please see Web interface installation page for information about Zabbix frontend installation wizard.

3 Installing Zabbix agent 2 loadable plugins

Installing Zabbix agent 2 loadable plugins is only required if you want to monitor targets not covered by built-in plugins (e.g., MongoDB servers or clusters, PostgreSQL and its forks, etc.). See the full list of loadable plugins and built-in plugins.

To install from sources, first download and extract the loadable plugin source archive.

To compile the plugin, navigate to the extracted plugin directory and run make:

make

The plugin executable may be placed anywhere as long as it is loadable by Zabbix agent 2. Specify the path to the plugin binary in the plugin configuration file, e.g. in postgresql.conf for the PostgreSQL plugin:

Plugins.PostgreSQL.System.Path=/path/to/executable/zabbix-agent2-plugin-postgresql

The path to the plugin configuration file must be specified in the Include parameter of the Zabbix agent 2 configuration file:

Include=/path/to/plugin/configuration/file/postgresql.conf

Proceed to setup for more details on configuring plugins.

Loadable plugins provided by Zabbix use simple makefiles with the following build targets:

• make - build the plugin
• make clean - delete all files that are created by building the plugin
• make check - run self-tests (requires a real monitoring target, e.g., a PostgreSQL database)
• make style - check Go code style with golangci-lint
• make format - format Go code with go fmt
• make dist - create a source archive including all dependencies

4 Installing Java gateway

It is required to install Java gateway only if you want to monitor JMX applications. Java gateway is lightweight and does not require a database.

To install from sources, first download and extract the source archive.

To compile Java gateway, run the ./configure script with --enable-java option. It is recommended that you specify the --prefix option to request installation path other than the default /usr/local, because installing Java gateway will create a whole directory tree, not just a single executable.

./configure --enable-java --prefix=$PREFIX

To compile and package Java gateway into a JAR file, run make. Note that for this step you will need javac and jar executables in your path.

make

Now you have a zabbix-java-gateway-$VERSION.jar file in src/zabbix_java/bin. If you are comfortable with running Java gateway from src/zabbix_java in the distribution directory, then you can proceed to instructions for configuring and running Java gateway. Otherwise, make sure you have enough privileges and run make install.

make install

Proceed to setup for more details on configuring and running Java gateway.

5 Installing Zabbix web service

Installing Zabbix web service is only required if you want to use scheduled reports.

To install from sources, first download and extract the source archive.

To compile Zabbix web service, run the ./configure script with --enable-webservice option.

Run zabbix_web_service on the machine, where the web service is installed:

zabbix_web_service

Proceed to setup for more details on configuring Scheduled reports generation.

6 Getting Zabbix source code

There are several ways of getting Zabbix source code:

• You can download the released stable versions from the official Zabbix website
• You can download nightly builds from the official Zabbix website developer page
• You can get the latest development version from the Git source code repository system:
  • The primary location of the full repository is at https://git.zabbix.com/scm/zbx/zabbix.git
  • Master and supported releases are also mirrored to Github at https://github.com/zabbix/zabbix

A Git client must be installed to clone the repository. The official commandline Git client package is commonly called git in distributions. To install, for example, on Debian/Ubuntu, run:

sudo apt-get update
       sudo apt-get install git

To grab all Zabbix source, change to the directory you want to place the code in and execute:

git clone https://git.zabbix.com/scm/zbx/zabbix.git

7 Compilation issues

These are the known issues regarding Zabbix compilation from sources. For all other cases, see the Known issues page.

Library in a non-standard location

Zabbix allows you to specify a library located in a non-standard location. In the example below, Zabbix will run curl-config from the specified non-standard location and use its output to determine the correct libcurl to use.

$ ./configure --enable-server --with-mysql --with-libcurl=/usr/local/bin/curl-config

This will work if it is the only libcurl installed in the system, but might not if there is another libcurl installed in a standard location (by the package manager, for example). Such is the case when you need a newer version of the library for Zabbix and the older one for other applications.

Therefore, specifying a component in a non-standard location will not always work when the same component also exists in a standard location.

For example, if you use a newer libcurl installed in /usr/local with the libcurl package still installed, Zabbix might pick up the wrong one and compilation will fail:

usr/bin/ld: ../../src/libs/zbxhttp/libzbxhttp.a(http.o): in function 'zbx_http_convert_to_utf8':
       /tmp/zabbix-master/src/libs/zbxhttp/http.c:957: undefined reference to 'curl_easy_header'
       collect2: error: ld returned 1 exit status

Here, the function curl_easy_header() is not available in the older /usr/lib/x86_64-linux-gnu/libcurl.so, but is available in the newer /usr/local/lib/libcurl.so.

The problem lies with the order of linker flags, and one solution is to specify the full path to the library in an LDFLAGS variable:

$ LDFLAGS="-Wl,--no-as-needed /usr/local/lib/libcurl.so" ./configure --enable-server --with-mysql --with-libcurl=/usr/local/bin/curl-config

Note the -Wl,--no-as-needed option which might be needed on some systems (see also: default linking options on Debian-based systems).

Stack size too small on some systems

If Zabbix crashes or freezes due to stack overflows, you can increase the per-thread stack size using the --with-stacksize option when configuring the sources. This issue may occur on systems with low default thread stack limits, especially during preprocessing, where multiple threads are created.

The following example sets the stack size to 512 KB per thread:

./configure --enable-server --with-mysql --with-stacksize=512

You can check the system thread stack limits at runtime using the ulimit -s command on Linux-based systems.