See for the official Zabbix site.


Jump to: navigation, search

Support of internal checks for proxies


Status: Stable, v1.2

Owner: Alexei


Currently there is no easy way to monitor health of Zabbix proxy.

Zabbix proxy should support internal items for monitoring proxy-related performance and availability metrics.


The following existing server-side internal items will be supported for proxies:

 zabbix[proxy_history]                  # number of values in the buffer for sending to server
 zabbix[items]                          # number of items
 zabbix[host,...]                       # host availability
 zabbix[hosts]                          # number of monitored hosts
 zabbix[items_unsupported]              # number of unsupported items
 zabbix[boottime]                       # proxy boot time
 zabbix[process,<type>,<mode>,<state>]  # same as for server, see notes below
 zabbix[queue,<from>,<to>]              # same as for server
 zabbix[rcache,<cache>,<mode>]          # same as for server
 zabbix[wcache,<cache>,<mode>]          # same as for server, but <cache> == 'trends' will not be supported by proxies

zabbix[process,<type>,<mode>,<state>] will support the following values for 'process' parameter for proxies: configuration syncer, discoverer, history syncer, housekeeper, http poller, icmp pinger, ipmi poller, java poller, poller, self-monitoring, snmp trapper, trapper, unreachable poller. These option will not be supported by proxies: alerter, db watchdog, escalator, node watcher, proxy poller, timer.

Also two new options will be introduced:

  • data sender
  • heartbeat sender

If the host is monitored directly by the server, internal items will return data about Zabbix server. If a host is monitored by a proxy, internal items will return data about Zabbix proxy and will be calculated by proxy.

Database changes

  • Add template for monitoring of proxies 'Template App Zabbix Proxy'
  • help_items should be updated including database patch


  • What's new
  • Upgrade notes must mention that such items will now return something else if a host is monitored by a proxy
  • Zabbix Manual
    • Update list of supported internal items

Also discussed

  • Possibility of sending internal metrics as part of heart beat messages
  • Prioritization for processing of items related to internal monitoring. It could be implemented in the future.
  • Use of proxy name as one of parameters for the internal checks. Rejected since it would introduce performance issues when filtering internal items related to a particular proxy.


  • v1.1
    • zabbix[history_proxy] renamed to zabbix[proxy_history]
  • v1.2
    • added zabbix[host,...] to the list of supported items by proxies