How does zabbix exactly count available RAM on Linux ?

[airocat]

Hello All!

I want to monitor amount of available RAM on my Linux server. But I don't completely understand how Zabbix gets this value.

Server OS - Debian 8, x86_64
Zabbix version - 3.0

According to the docs:

https://www.zabbix.com/documentation/2.4/manual/appendix/items/vm.memory.size_params

From Platform-specific notes we can see:

On Linux available is free + buffers + cached (There is no such information in 3.0 section, but I suppose it should be the same)


I can see different picture in my case:

#zabbix_get -s host -k 'vm.memory.size[free]'
252035072
#zabbix_get -s host -k 'vm.memory.size[buffers]'
1649950720
#zabbix_get -s host -k 'vm.memory.size[cached]'
6410240000

The sum off free buffers and cached is 8312225792.

But:

# zabbix_get -s host -k 'vm.memory.size[available]'
10890870784

So, how really this value was calculated ?

Thanks in advance

[guzzijason]

Perhaps it varies based on OS?

On an Ubuntu 14.04 host, the output of free -m looks like this:

# free -m
             total       used       free     shared    buffers     cached
Mem:        257837       7737     250099         16        279       4927
-/+ buffers/cache:       2530     255306
Swap:            0          0          0

and in this case, the "available" memory reported by zabbix seems to correlate with ( free + buffers + cached ) / 1024 .

Howerver, on a CentOS 7 host, free -m looks like this:

# free -m
              total        used        free      shared  buff/cache   available
Mem:           7822        5288         575         247        1958        1998
Swap:          9215           0        9215

... and the "available" memory reported by zabbix seems to line up the the available memory shown in the free output (divided by 1024).

__Jason