I've encountered 3 crashes of the zabbix_server. The stack trace shows that the last calls were from the LLD worker, but in the log file I see this:

[file:dbconfig.c,line:109] __zbx_shmem_malloc(): out of memory (requested 1656471 bytes)
[file:dbconfig.c,line:109] __zbx_shmem_malloc(): please increase CacheSize configuration paramater

I can confirm that the CacheSize, according to the built-in Zabbx server internal metric "Configuration cache, % used", was 49% utilized at the time of the crash. 

In the zabbix_server.conf the CacheSize parameter is set to 20G. A minute before time of crash the resident set size was 19.0GB. I can also see that over time the resident set size has been slowly increasing, indicating a possible memory leak?

Let me know if you need additional details about the call stack in the LLD worker at time of crash.