Zabbix proxy is a process that may collect monitoring data from one or more monitored devices and send the information to the Zabbix server, essentially working on behalf of the server. All collected data is buffered locally and then transferred to the Zabbix server the proxy belongs to.
Deploying a proxy is optional, but may be very beneficial to distribute the load of a single Zabbix server. If only proxies collect data, processing on the server becomes less CPU and disk I/O hungry.
A Zabbix proxy is the ideal solution for centralized monitoring of remote locations, branches and networks with no local administrators.
Zabbix proxy requires a separate database.
Note that databases supported with Zabbix proxy are SQLite, MySQL and PostgreSQL.
See also: Using proxies in a distributed environment
Zabbix proxy runs as a daemon process. The proxy can be started by executing:
This will work on most of GNU/Linux systems. On other systems you may need to run:
Similarly, for stopping/restarting/viewing status of Zabbix proxy, use the following commands:
If the above does not work you have to start it manually. Find the path to the zabbix_proxy binary and execute:
You can use the following command-line parameters with Zabbix proxy:
-c --config <file> path to the configuration file
-f --foreground run Zabbix proxy in foreground
-R --runtime-control <option> perform administrative functions
-T --test-config validate configuration file and exit
-h --help give this help
-V --version display version number
Examples of running Zabbix proxy with command-line parameters:
Runtime control options:
Option | Description | Target |
---|---|---|
config_cache_reload | Reload configuration cache. Ignored if cache is being currently loaded. Active Zabbix proxy will connect to the Zabbix server and request configuration data. Passive Zabbix proxy will request configuration data from Zabbix server the next time when the server connects to the proxy. |
|
diaginfo[=<section>] | Gather diagnostic information in the proxy log file. | historycache - history cache statistics preprocessing - preprocessing manager statistics locks - list of mutexes (is empty on BSD systems) |
snmp_cache_reload | Reload SNMP cache, clear the SNMP properties (engine time, engine boots, engine id, credentials) for all hosts. | |
housekeeper_execute | Start the housekeeping procedure. Ignored if the housekeeping procedure is currently in progress. | |
log_level_increase[=<target>] | Increase log level, affects all processes if target is not specified. Not supported on BSD systems. |
process type - All processes of specified type (e.g., poller) See all proxy process types. process type,N - Process type and number (e.g., poller,3) pid - Process identifier (1 to 65535). For larger values specify target as 'process type,N'. |
log_level_decrease[=<target>] | Decrease log level, affects all processes if target is not specified. Not supported on BSD systems. |
|
prof_enable[=<target>] | Enable profiling. Affects all processes if target is not specified. Enabled profiling provides details of all rwlocks/mutexes by function name. |
process type - All processes of specified type (e.g., history syncer) See all proxy process types. process type,N - Process type and number (e.g., history syncer,1) pid - Process identifier (1 to 65535). For larger values specify target as 'process type,N'. scope - rwlock , mutex , processing can be used with the process type and number (e.g., history syncer,1,processing) or all processes of type (e.g., history syncer,rwlock) |
prof_disable[=<target>] | Disable profiling. Affects all processes if target is not specified. |
process type - All processes of specified type (e.g., history syncer) See all proxy process types. process type,N - Process type and number (e.g., history syncer,1) pid - Process identifier (1 to 65535). For larger values specify target as 'process type,N'. |
Example of using runtime control to reload the proxy configuration cache:
Examples of using runtime control to gather diagnostic information:
# Gather all available diagnostic information in the proxy log file:
zabbix_proxy -R diaginfo
# Gather history cache statistics in the proxy log file:
zabbix_proxy -R diaginfo=historycache
Example of using runtime control to reload the SNMP cache:
Example of using runtime control to trigger execution of housekeeper
Examples of using runtime control to change log level:
# Increase log level of all processes:
zabbix_proxy -c /usr/local/etc/zabbix_proxy.conf -R log_level_increase
# Increase log level of second poller process:
zabbix_proxy -c /usr/local/etc/zabbix_proxy.conf -R log_level_increase=poller,2
# Increase log level of process with PID 1234:
zabbix_proxy -c /usr/local/etc/zabbix_proxy.conf -R log_level_increase=1234
# Decrease log level of all http poller processes:
zabbix_proxy -c /usr/local/etc/zabbix_proxy.conf -R log_level_decrease="http poller"
Zabbix proxy is designed to run as a non-root user. It will run as whatever non-root user it is started as. So you can run proxy as any non-root user without any issues.
If you will try to run it as 'root', it will switch to a hardcoded 'zabbix' user, which must be present on your system. You can only run proxy as 'root' if you modify the 'AllowRoot' parameter in the proxy configuration file accordingly.
See the configuration file options for details on configuring zabbix_proxy.
agent poller
- asynchronous poller process for passive checks with a worker threadavailability manager
- process for host availability updatesconfiguration syncer
- process for managing in-memory cache of configuration datadata sender
- proxy data senderdiscovery manager
- manager process for discovery of devicesdiscovery worker
- process for handling discovery tasks from the discovery managerhistory syncer
- history DB writerhousekeeper
- process for removal of old historical datahttp agent poller
- asynchronous poller process for HTTP checks with a worker threadhttp poller
- web monitoring pollericmp pinger
- poller for icmpping checksipmi manager
- IPMI poller manageripmi poller
- poller for IPMI checksjava poller
- poller for Java checksodbc poller
- poller for ODBC checkspoller
- normal poller for passive checkspreprocessing manager
- manager of preprocessing tasks with preprocessing worker threadspreprocessing worker
- thread for data preprocessingself-monitoring
- process for collecting internal server statisticssnmp poller
- asynchronous poller process for SNMP checks with a worker thread (walk[OID]
and get[OID]
items only)snmp trapper
- trapper for SNMP trapstask manager
- process for remote execution of tasks requested by other components (e.g. close problem, acknowledge problem, check item value now, remote command functionality)trapper
- trapper for active checks, traps, proxy communicationunreachable poller
- poller for unreachable devicesvmware collector
- VMware data collector responsible for data gathering from VMware servicesThe proxy log file can be used to observe these process types.
Various types of Zabbix proxy processes can be monitored using the zabbix[process,<type>,<mode>,<state>] internal item.
Zabbix proxy runs on the same list of supported platforms as Zabbix server.
The memory buffer allows to store new data (item values, network discovery, host autoregistration) in the buffer and upload to Zabbix server without accessing the database. The memory buffer has been introduced for the proxy since Zabbix 7.0.
In installations before Zabbix 7.0 the collected data was stored in the database before uploading to Zabbix server. For these installations this remains the default behavior after upgrading to Zabbix 7.0.
For optimized performance, it is recommended to configure the use of memory buffer on the proxy. This is possible by modifying the value of ProxyBufferMode from "disk" (hardcoded default for existing installations) to "hybrid" (recommended) or "memory". It is also required to set the memory buffer size (ProxyMemoryBufferSize parameter).
In hybrid mode the buffer is protected from data loss by flushing unsent data to the database if the proxy is stopped, the buffer is full or data too old. When all values have been flushed into database, the proxy goes back to using memory buffer.
In memory mode, the memory buffer will be used, however, there is no protection against data loss. If the proxy is stopped, or the memory gets overfilled, the unsent data will be dropped.
The hybrid mode (ProxyBufferMode=hybrid) is applied to all new installations since Zabbix 7.0.
Additional parameters such as ProxyMemoryBufferSize and ProxyMemoryBufferAge define the memory buffer size and the maximum age of data in the buffer, respectively.
Note that with conflicting configuration the proxy will print an error and fail to start, for example, if:
Note that the proxy requires a UTF-8 locale so that some textual items can be interpreted correctly. Most modern Unix-like systems have a UTF-8 locale as default, however, there are some systems where that may need to be set specifically.