Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I've setup Zabbix on a small VPS recently and got it to monitor three other VPS's. Already we've got about 9.8 incoming values per second and the load on the Zabbix server are already topping. It's got 256M of memory along with APC as opcode cache to speed up PHP.

Is it possible to reduce the granularity, ie lower the incoming values to offload our Zabbix server as we're not monitoring misson-critical stuff?

share|improve this question
up vote 3 down vote accepted

you got 2 ways of doing this:

edit the templates used to increase the "interval" of checks (there is a mass update ability but use it sparingly), this means checks are done less often and thus should spare some capacity

on each vps, disable the checks you know you dont need, less checks means less work to do

also try editing your zabbix server and agent config on the monitoring vps to lower the number of processes it starts, to something like 2 agents and 4 servers


for agent config


for server config

i have to point out that you are running on what is effectively the minimum requirements for zabbix server so you will have to reduce the amount of data you are processing if you want to add more, because zabbix is reliant on innodb with mysql you are pushing resource limits to start with, doubling ram would be a safer and more reliable option

share|improve this answer
Thanks a lot Anthony! Will try these changes out. I know that we're living on the edge with these resources, but as a startup we're flipping every coin to get the most back :) – Industrial Aug 19 '11 at 8:47
would something like pingdom not be a better fit then? – anthonysomerset Aug 19 '11 at 9:11
We'll be using Pingdom too, but only to ensure uptime from a third-party. ZABBIX is great to monitor certain applications/services and their impact on load – Industrial Aug 19 '11 at 9:52
Reduce the frequency at which items are being gathered, the defaults are quite resource-hungry. Use mostly active items instead of passive, it's better for the load distribution as well. Alternatively, with that few servers, you might want to look at hosted monitoring. is hosted zabbix, which might be saving you a lot of hassle. Disclaimer: I'm the founder of tribily ;) – Walter Heck Aug 21 '11 at 11:06
Reducing the number of trappers and pollers can cause you queue to back up and can lead to gaps in graphs. You can most often leave the trappers/pollers as they are by default as their resource footprint is very small. – Red Tux Oct 3 '11 at 16:58

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.