NPM 10--monitoring of ESX 4 Servers

We have recently upgraded from Orion NPM 9.5 to NPM 10, in order to monitor our VMWare ESX servers. Our current environment has a mix of ESX 3.5 and ESX 4 machines, and ESX 3.5 monitoring appears to work correctly. However, after following all applicable instructions in the NPM 10 documentation for configuring SNMP on ESX 4.0, the node details page is not displaying any CPU utilization or memory usage data for the host. Individual VM data is available in the appropriate tables, but overall host CPU and memory radial gauges are absent for our ESX 4.0 servers (note that they are present for our ESX 3.5 servers). Is this a bug, perhaps a misconfiguration, or working as intended?

Find more posts tagged with

Accepted answers

WBolten

Ahum, consider this one a newby error... For some reason I though the tick for the ESX poller was set, but it was not...

Sorry

All comments

lchance

we are having many issues stablizing SNMP with ESX 4 and seeing the CPU & MEMORY gauges also. we don't get NIC or DISK VOLUMES either.

when we do get SNMP working it spikes the ESX system's CPU and we have to stop the SNMP process.

we do have a high pri case opened with VMware.

we really need to monitor the hardware on our systems - most are Dell - and we're trying to get SNMP to stay stable to try and use Dell's OpenManage for ESX.

(NPM 10)

lchance

this may be a real stupid question (this i can do well) but i notice our servers are 64bit Linux - is this any concern to work with Orions NPM 10 ESX API?

Linux myhostname 2.6.18-128.ESX #1 Fri Apr 10 00:08:17 PDT 2009 x86_64

bshopp

Please post back what you hear from VMWare.

How did you setup SNMP, did you look at the Admin guide or the doc here?

FormerMember

We followed the instructions listed in the Orion NPM 10 admin guide.

Donald_Francis

Lchance, that sounds scarily similar to the issue I had on a couple of Cisco chassis's where I had to stop SNMP for the CPU to not peg out after getting "snmp queue full" errors from the chassis.

lchance

We've followed the Admin Guide.

But are we supposed to be seeing the ESX Host CPU - MEMORY - Volumes - NIC? We do not.

And if we use only the API we see the Guests Systems but if we include SNMP we don't.

bshopp

You can use both SNMP and the API at the same time.

Via API we gather the list of guest and the stacked ESX charts that show guest cpu, memory and network utilization.

Via SNMP we gather volume, interface stats and the ESX host CPU/Memory.

WBolten

Yesterday I installed version 10 and added two 3.5 ESX hosts. For both hosts I get CPU and Memory information from the hosts themselves but not from the VM's running on them. Simple see the message that there is no data being collected for this.

In the FP I see a comment that 3.5 works well. How did you manage to get information from 3.5?

inrouted

We have a similar problem. The CPU dials appear, but always report the CPU at 100%. Examination of the box proves otherwise though.

FormerMember

Yesterday I installed version 10 and added two 3.5 ESX hosts. For both hosts I get CPU and Memory information from the hosts themselves but not from the VM's running on them. Simple see the message that there is no data being collected for this.
In the FP I see a comment that 3.5 works well. How did you manage to get information from 3.5?

We followed the instructions given in the Orion NPM 10 admin guide regarding configuration and adding of the ESX 3.5 servers in Orion and have had no problems since. If you are missing the data from specific VMs, it sounds as if you are not receiving the API data correctly, perhaps?

WBolten

Ahum, consider this one a newby error... For some reason I though the tick for the ESX poller was set, but it was not...

Sorry

bigdaddy33

Just wanted to mention that we are seeing the same behavior on our ESX4 servers. We had a fresh install of NPM version 10 and our 3.5 ESX boxes work perfectly while our 4.0's have this issue. We too followed the instructions with the same results