Hi all
I've written a monitoring Perl script that runs resxtop against my ESXi 4.0 server, grabs out the (for me important figures from the overwhelming amount of data and pumps it into my monitoring tool (Zabbix). So far so good, I seem to get reasonable numbers for CPU, memory and network, but the per-VM disk IO numbers are (almost) alway 0.
For example: I run an intensive IO test (iozone) in one of my Linux VMs. When I attach resxtop from my monitoring machine to the ESXi host, I can see IO rates (CMDS/s, etc.) in the range of 800 - 1400, matching my expectations. At the same time from the same monitoring machine, I capture resxtop statistics in batchmode, e.g.
# resxtop --server xxx --username zabbix -b -n 3 > /tmp/xxx
After that, I find out that the disk IO stats for the VM under test are in columns 10029-10033. Grabbing them shows me:
# cut -d, -f10029-10033 /tmp/xxx "\\xxx\Physical Disk(myvm)\Commands/sec","\\xxx\Physical Disk(myvm)\Reads/sec","\\xxx\Physical Disk(myvm)\Writes/sec","\\xxx\Physical Disk(myvm)\MBytes Read/sec","\\xxx\Physical Disk(myvm)\MBytes Written/sec" "3570224654039742567.75","3570224654042983646.75","3570224654042480460.00","3404831467986.96","3404831503500.18" "0.00","0.00","0.00","0.00","0.00" "0.00","0.00","0.00","0.00","0.00"
First thing to note is that the first line always cotains huge numbers (ok, easy enough to strip those away). But troubling is that almost all the following values are 0. For all VMs, on different ESXi hosts. I DO see correct numbers in interactive resxtop, though!
I can't make any sense of that. Am I doing something wrong here? Or am I using the wrong metrics (columns) - there are so many to choose from
Any help appreciated!
CU, Joe
Your post has been moved to the vCLI forum.
Dave
VMware Communities User Moderator
Now available - vSphere Quick Start Guide
Do you have a system or PCI card working with VMDirectPath? Submit your specs to the Unofficial VMDirectPath HCL.
I'm getting the same strange values for VM disk I/O using resxtop in batch mode, within vMA too.
These kind of values comes from ESX 4 only. The same resxtop in batch mode against ESX 3.5 gives realistic values.
So I suppose it could be a bug on VMware ESX Management services on ESX4 but I didn't found any article about this in KB.
Anybody solved this?
thanks
angelo
This seems to be a bug in resxtop. I changed my script to use esxtop on the ESXi server via SSH, and now I get values for VM disk metrics that are not 0. Still have to check if they make sense, but they are not 0 anymore.
So this seems to be a vCLI issue. I'm using vCLI version 1.6 (I guess) on 64-bit Linux (SLES 10SP2). Server is VMware ESXi 4.0.0 build-208167
CU, Joe
root@xxx /usr/lib/vmware-vcli# /usr/bin/esxcli --version Script 'esxcli' version: 4.0 root@xxx /usr/lib/vmware-vcli# uname -a Linux xxx 2.6.16.60-0.42.5-smp #1 SMP Mon Aug 24 09:41:41 UTC 2009 x86_64 x86_64 x86_64 GNU/Linux root@xxx /usr/lib/vmware-vcli# esxcfg-nics --version VI Perl Toolkit version: 1.6 Script 'esxcfg-nics' version: 3.5 Update 2
If you believe this is in fact a bug, I would recommend filing an SR with VMware Support to report this bug and hopefully it'll get resolved if not already.
=========================================================================
William Lam
VMware vExpert 2009
VMware ESX/ESXi scripts and resources at:
VMware Code Central - Scripts/Sample code for Developers and Administrators
If you find this information useful, please award points for "correct" or "helpful".