calculations in items and/or graphs

[Aaron]

I have values for total memory and free memory. I need to make a graph of memory used. It seems that I cannot graph a calculation since my only options are from a drop down box of existing items. I have been trying to define an item that is basically (total - free) to give me a "memory used" item value that I can graph. So far I haven't managed to get anywhere.

Is this even possible?

I am currently considering looking at the agent and hacking to add a new "used" mode to vm.memory.size. Any opinion on that? I would prefer limit my customization to the interface itself because it would make future upgrades and such easier. That and my programming skills are lousy.

[Aaron]

I created a UserParameter entry that calls a script for this so I have the amount of memory used stored as a variable in the database.

[vrtareg]

Hi

Can you share your script content ?
I want to do the same to see some parameters on same graph.
For example memory usage and CPU utilization in percents to determine high loaded servers.

Thanks
Areg

[lcondado]

This is an example of memory calculation usage (percent) in a linux server with 4Gb of memory

vmstat 1 2 | /bin/gawk '{if (NR==4) print ((4000000-$5)*100)/4000000}'

in this case vmstat reports memory size in KB, however in HPUX the memory usage is reported in pages.

[achu_sab]

Can you specify how the script is implemented? I tried to add it to zabbix_agentd.conf as a user parameter but it doesn't work, pls describe where I have to add the script to get the memory usage, Also can you tell me where is the location of file in which zabbix_agentd keys are specified?

[SierraX]

Hi, is there another solution for this problem?
I have intigrate a programmline to send the a sum of 3 Memorys sizes.
UserParameter=real.free.mem, free -b|awk ‘$1==”-/+” { print $NF }’
and it works.
But this is a very dirty way. I mean if i have a Databaseserver, did i realy need to send 2 times the same Informations?

[lcondado]

Here is the script (for solaris):

cat mem_used.ksh
#!/bin/ksh

vmstat 1 2 | /bin/awk '{if (NR==4) print ((32000000-$5)*100)/32000000}'

And here is the agent file configuration line:

grep mem_used.ksh /etc/zabbix/zabbix_agentd.conf
UserParameter=CUST.MEM_PERC,/herramientas/monitoreo/zabbix/scripts/mem_used.ksh


For linux is reccomendable to subtract the cached memory:

cat mem_used.ksh
#!/bin/bash

vmstat 1 2 | /bin/gawk '{if (NR==4) print (((16400000-$6)-$4)*100)/16400000}'

[cbidwell]

Would you explain here how you implemented this into the zabbix fe configuration? I've been trying something similar to this but I only get 'no data'.