Less than 25% free in the value cache

[sbrews]

First - I am new to zabbix and know next to nothing about it (but I am expected to be an expert ).

I received this alert from the zabbix server this morning:

Trigger: Less than 25% free in the value cache Trigger status: PROBLEM Trigger severity: Average Trigger URL:
Trigger Hostgroup: Discovered Hosts, Linux DCW, Linux_Corp, Zabbix

Item values:

1. Zabbix value cache, % free (s5437calv:zabbix[vcache,buffer,pfree]): 24.99 % 2. *UNKNOWN* (*UNKNOWN*:*UNKNOWN*): *UNKNOWN* 3. *UNKNOWN* (*UNKNOWN*:*UNKNOWN*): *UNKNOWN*

Original event ID: 8227208

Searching around the system, I found the zabbix_server.conf file and see that ValueCacheSize is set to 24M.

What is the recommended way to resolve this alert? Do I increase the setting? Do I stop/clear something else?

As I mentioned above, I am new to zabbix, so what should be obvious/easy is not (yet).

[DmitryL]

Hello sbrews,

From the start:
Fresh installed Zabbix server process is fully functional and will work, however when your zabbix environment will grow you will have to tune your zabbix_server.conf file (increase count of process/cache) also don't forget to tune your database!

How to know what and when to increase?

From the box Zabbix has templates to monitor the health of zabbix server itself.
For Server this template is called: Template App Zabbix Server
Inside this template you are able to find graphs that will tell you current health of server.
Simply go to Monitoring -> Graphs and choose Zabbix server host
You are interested in graphs:
Zabbix server performance Here you will see real number processed by server each second and queue.
Zabbix internal process bussy
Zabbix gathering process bussy
Zabbix cache usage % free

If you see on the graphs, that some process is lets say more then 50% busy, it could be the time to schedule server restart. Same idea with cache.
All these parameters can be increased in /etc/zabbix_server.conf file (restart of server is required after changes are made)

How much to increase?
Hard to say. I would suggest to experiment. Increase parameter - look the graphs again. Still busy? Increase more.

NOTE: even if you have very powerfull server running Zabbix, it does not mean that you should increase all parameters 5 times. You will have to find that close to perfect number, when process is not busy and you won't have any processes without work.

Hope got this a little clearer to you.
Also recommendation to read documentation, you will find answers on many questions there!

[cesarsj]

If I increase CacheSize, do I really need to restart the zabbix_server process or can I simply run zabbix_server -R config_cache_reload?

Does this command tell Zabbix that I have increased CacheSize or just reloaded the cache?

[dimir]

Unfortunately, you really need to restart the whole Zabbix server. Reloading configuration cache has nothing to do with re-reading configuration file.
But don't forget to vote on a feature request: https://support.zabbix.com/browse/ZBXNEXT-3162

[cesarsj]

Understand. I had already voted for this request. I believe Zabbix will be much better when it can reload configurations without having to restart the server and especially export reports (I saw in a post that there are companies that don't migrate from Nagios to Zabbix because of this. What is the Zabbix team waiting for? Will Zabbix 5 come with these features?)