Server health analysis

[ebonweaver]

So I just found out about the missing template and got that imported and now I'm looking at all this wonderful data, with no clue what most of it is or if it's healthy. I suspect several of these processes that are always at 100% shouldn't be? Any evaluation and recommendation on these is appreciated!

[ebonweaver]

I should mention, the ones that show no data are shown in the items as not supported. I'm not sure I understand how they are not supported when they are in the template from zabbix.....

[Firm]

In case of unsupported processes it means that you have no items that are using them. 100% busy for particular zabbix process means you need to increase number of such processes by editing config file.

[Ed.M]

Hi!
Zabbix cache usage is OK in this case. You've got pretty good numbers, cache almost free, no problems here.

Now let's look to the first graph. There are problems with discoverer, pinger and unreachable poller. What can you do: disable discovery tasks you don't need. Maybe set update interval to longer time. If there is no chance to optimize discoverer process that way, then you need to increase process number in config file.
Pinger: if you do a lot of icmp checks - increase process number. But do not set maximum. Try increasing until you'll get approximately 40% of load. That would be really good.
Unreachable poller - please check if you have a lot of old unreachable hosts enabled for monitoring. Disable them. Those timeouts make real load to unreachable poller process. If not - again, increase process number, till its load drops to 40%.