Most data in Node Details view is flat out wrong!!

Question

Hi,

I've installed and have had running Orion for about a month now. My network is quite small....about 10 routers and a few hundred PCs and servers. I have enabled SNMP on as much as I can and now want to collect data and monitor devices. All devices that are SNMP enabled can have their system oid, etc read...but things get a little confusing later. In the Node details page of the web console for various nodes the CPU utilization is clearly wrong. It will say something like 6 or 7 percent and will stay at this value when I run a stress test on the machine jacking its CPU up to 50% for an hour or so. Memory Usage is not very reliable either. On one Windows system it's reporting 8% yet the Task Manager on that system show 500Mb out of 4GB being used (12%) and the 4GB includes Swap Space. Also interface traffic doesn't seem to show the correct data. It sits there at 0% even though I'm copying large file through a device's NIC and pushing the network traffic up to 35%. Am I doing something completely wrong????

Basically what I want to do is (reasonably) accurately monitor %CPU usage, %Memory usage, %Disk space, %Network Card Traffic and Status (up/down) of about 50 computers which are Windows XP or Linux. This shouldn't be so hard. Thanks for any help.

Jackie

AndrewShine · Answer

Jackie,

Your original post mentioned that window memory monitoring was wrong. I noticed recently that the percentage reported on the node memory gauge is seems to be an average of the physical memory usage and the pagefile usage. Might explain one of your discrepancies?

Regards

Andrew

vhcato · Answer

Torrie,

The only thing I can think of for the differences in values you're seeing is to make sure you're not comparing "current" with "average", or something of that nature. I haven't seen the symptoms you describe in our environment. Can you perhaps zip up some screen shots of the inconsistencies you're seeing, attach them to this post (or create a new one of your own), and let's see if anyone can pick out what might be happening?

And can you do the same for the alerting issue?

Maybe you already have, and I'm just not putting it together... :(

torrielw · Answer

DUDE....I know your pain. My Director has doted on Solarwinds simplicity, reliability and accuracy versuses other more expensive monitoring tools for over a year. And today I still can't get it to alarm accurately on simple things like percent utilization. I mean how does the web console show high utilization but the Alerting engine can't (that includes both versions 7.85 and 8.1)? Recently I made a copy of Current Top Ten and filtered it for specific devices .....but Node Detail differs from Top 10 Current Response Time. How is that??  Where exactly is Orion getting it's data from?  And why does one web page report one thing and another something totally different?

My NOC team members have decided to ignore most of the information they see (like interfaces with 192% Receive utilization) I'm still hacking away at it though and really dependent on the more experience users in this forum.

You are not alone.