Please increase configuration cache

[shahib02]

Last week I have installed zabbix 3.4 on Ubuntu 16.04. Initially everything was working fine. After I add near 15 switches using snmp. Zabbix server is not running. I check log it shows configuration cache out of memory. I modify zabbix_server.conf. Conf file to increase cache size to 8G. Still not working. Kindly advice me.

[kloczek]

Check logs again.

[shahib02]

Dear Bro

This is problem with auto discovery rule for interface. i disable

Network Interfaces Discovery

problem solved. but my question is why configuration cache is become too low for network interface discovery. i already give 8G for configuration cache.

[shahib02]

Kindly give any suggestion because I need to connect near 20 Cisco switch. If I connect near 10 switches, configuration cache go to low

[kloczek]

With 15 switches you've already saturated 8GB of config cache.
What you are monitoring on those switches?

[shahib02]

I use this template for all switches (Template Module Cisco Inventory SNMPv2, Template Module Generic SNMPv2), Template Module Interfaces Simple SNMPv2.
if i use this template (Template Module Interfaces Simple SNMPv2) . it automaticall discover more item near 2000.

4507_Switch

Applications 4

Items 2089

Triggers 941

Graphs 228

Discovery 2

i disable near 1500 item. but still configuration cache not reduced.

My set up is Ubuntu 16.04 & Zabbix server 3.4 on VMWare machine.

[kloczek]

Really it does't matter which one version of the distribution and zabbix you are using because using 8GB of cfg cache is only related to how many items you have already registered in you zabbix.

8GB of config cache it is a lot. Especially if you will put this in context of the only 15 switches.
How many items per such switch you have (ruffly) and how many items in whole zabbix you have now registered?
With 8GB cfg cache already saturated it may mean that you've already may have something like 8mln items (usually item in cache takes 1KB). It is more than 500k items per device. If nothing more is monitored over this zabbix it may mean that literally you have registered all possible OIDs reported by those devices.
So far you did not provide to much details so above may be wrong but if you are worry about so big cfg cache usage and my calculations are enough close to what you really have I would focus more on what exactly you are trying to monitor on those switches.

[shahib02]

Thanks brother for your support

Now i solve this issue because of when i change configuration cache in zabbix_server.conf file before i didn't remove # symbol. now i remove this symbol all is good.