This documents decribes how ZABBIX server handles timeout situations and network errors.
Note that ZABBIX equally treats timeouts and network errors.
ZABBIX poller processes periodically (according to configuration of items) queries ZABBIX and
SNMP agents for up-to-date performance and availability information. Normally, if no timeout happens, it connects to the
agent, asks for needed information, receives the data, disconnects and processes the received data.
But what happens in case of timeout?
First timeout
ZABBIX writes Host [<hostname>]: first network error, wait for 15 seconds to
server's log file and will try to connect to the host only after UnreachableDelay seconds, where UnreachableDelay = 15 by default.
So, normally the host will be checked after 15 seconds.
Next timeouts
After 15 seconds, ZABBIX server will try to connect to the monitored device once again.
In case of timeout and if a host was unavailable for more than UnreachablePeriod, 45 seconds by default, the host will be checked after UnavailableDelay seconds. Host status will be changed to Unreachable.
UnavailableDelay is 60 seconds by default.
ZABBIX writes Host [<hostname>]: another network error, wait for 15 seconds to the log file.
If more than UnreachablePeriod seconds passed since first timeout, the host status will be changed
to UNAVAILABLE and it will be checked again each UnavailablePeriod seconds, 60 seconds by default.
Important
One of ZABBIX pollers is dedicated to processing of unreachable and unavailable hosts. The poller does not process normal hosts.
| 
Database structure