Why sometimes zabbix server crashes?

Question

Zabbix

Sometimes Zabbix server crashes , the period between crashes varies from 1 week to a month.
The server crashes for about 2 hours . At the time of the last crash, I accidentally ended up on a server and decided to collect information about the crash, and as a result, I got upset while the GUI was saying "Zabbix server is not running: the information displayed may not be current":

1. The main process works like all pullers, trappers and the rest.
   
2. There is no information in the logs about the operation of the server itself, there are only records of timeouts from some agents during the collection of metrics, given that monitoring is done on vps in the states, and controlled servers in Russia are, in principle, normal, nothing out of the ordinary.
   

In general, messages periodically slip through the logs: cannot send list of active checks to "xxx.xxx.xxx.xxx": host [NNNNNN] not found , which hints that the server is not available, but why and which part of it is not available and how treat.
In the meantime, I set DebugLevel = 3 and wait for the next crash.
I would like to hear in which direction to dig?
UPD:
It falls, the fiction stops collecting statistics and notifying about events, and after it rises, it starts pouring out triggers, due to the lack of information for a period of about 2 hours.
UPD2:
today 2018-02-20 at 15:00 again there was a fall

confuses the

UPD3 processor usage schedule:
Maybe a coincidence, but I don’t think ... in the zabbix server log at 15:24:43 the first entry appeared "Lost connection to MySQL server during query ..." and the last such entry for the whole day was at 17:03:19 all such there were 32 entries during the fall. But the problems started at 15:00.
In the evening I'll go deeper into the logs ..

Answer 1

Check syslog and dmesg