Utilization of preprocessing manager processes over 75%

[zabbix1025]

Zabbix currently monitors 500 servers, and I recently upgraded from 5.0.1 to version 6.0.3. I often need to restart the Zabbix Server process to resolve the "Preprocessing Manager Processes over 75%" problem. In version 5.0.1, the number of preprocessing worker processes was stable at 30. After upgrading to version 6.0.3, the number increased to 500, but still had the same problem.


the docker-compose files：

version: "3"
services:
zabbix-server:
image：zabbix/zabbix-server-mysql:6.0.3-ubuntu
container_name: zabbix-server
ulimits:
nproc: 65535
nofile:
soft: 65535
hard: 65535
memlock:
soft: -1
hard: -1
privileged: true
user: zabbix
env_file:
- /data/zabbix/zabbix_server.conf
cap_add:
- ALL
restart: always
network_mode: host

CONFIGURATION


ZBX_TIMEOUT=30
ZBX_LISTENPORT=10051
ZBX_STARTPOLLERS=95
ZBX_IPMIPOLLERS=10
ZBX_STARTPREPROCESSORS=500
ZBX_STARTPOLLERSUNREACHABLE=45
ZBX_STARTTRAPPERS=25
ZBX_STARTPINGERS=25
ZBX_STARTDISCOVERERS=35
ZBX_STARTHISTORYPOLLERS=130
ZBX_STARTHTTPPOLLERS=25
ZBX_STARTTIMERS=15
ZBX_STARTESCALATORS=95
ZBX_STARTALERTERS=35
ZBX_STARTLLDPROCESSORS=35
ZBX_HOUSEKEEPINGFREQUENCY=1
ZBX_MAXHOUSEKEEPERDELETE=10000
ZBX_PROBLEMHOUSEKEEPINGFREQUENCY=60
ZBX_CACHESIZE=1024M
ZBX_CACHEUPDATEFREQUENCY=10
ZBX_STARTDBSYNCERS=50
ZBX_HISTORYCACHESIZE=1024M
ZBX_HISTORYINDEXCACHESIZE=1024M
ZBX_HISTORYSTORAGEDATEINDEX=0
ZBX_TRENDCACHESIZE=512M
ZBX_TRENDFUNCTIONCACHESIZE=512M
ZBX_VALUECACHESIZE=1024M
ZBX_TRAPPERTIMEOUT=300
ZBX_UNREACHABLEPERIOD=45
ZBX_UNAVAILABLEDELAY=60
ZBX_UNREACHABLEDELAY=15
ZBX_LOGSLOWQUERIES=3000
ZBX_STARTPROXYPOLLERS=50
ZBX_PROXYCONFIGFREQUENCY=60
ZBX_PROXYDATAFREQUENCY=120
ZBX_STARTREPORTWRITERS=5

[cyber]

You definitely start up too many processes for just 500 servers...
500 preprocessors? Even 5 is enough there... rest of them START... options - divide by 10 or more..:P

[zabbix1025]

You definitely start up too many processes for just 500 servers...
500 preprocessors? Even 5 is enough there... rest of them START... options - divide by 10 or more..:P

I see the documentation that increases the number of processes when this alarm occurs. But MY attempts to increase or decrease the number of processes failed to solve the problem.