I have Orion monitoring a bunch of RHEL 5 servers. I've been looking around for this info, but I'm having a hard time wading through a bunch of posts - many of which are a few years old. So, I thought I'd ask my questions here.
By default, how does Orion handle CPU and memory usage for linux boxes? Specifically, how does it handle the CPU usage for servers with multiple cores? I have one server that has 4 quad core processors. I got an alert last night that the CPU spiked to 100%. In digging deeper into what Orion saw, there were actually 2 spikes - 10 minutes apart from each other. Since there are 16 cores on this server, does that alert mean that all of the cores went to 100% or just 1 core? A coworker and I went through the logs for the application that runs on that server, and it was hardly doing anything at all when these spikes occurred. In fact when looking at the CPU chart in Orion for the last 7 days, these 2 spikes are all that's there. The rest of the time the CPU is practically just idling.
On the memory side, Orion seems to be combining the memory that is actually being used with the memory that is cached - providing a false high report. Orion says that over 90% of the RAM is being used, but from looking at the server, it's not paging at all.
Thanks in advance for any help. I'm not at all experienced with Orion or even SNMP in general.
