Virtual Machine Network Usage more than 100%. Why?

[olegarr]

Hello:

Hope someone will explain me the odd alerts that I am receiving...

I have VI2 with ESX servers 3. I set up some alarms and one of them supposed to send me email if "Virtual Machine Network Usage" will exceed 75% (warning) and 90% (alert). It's working fine, however once in a while I am receiving alerts with message that my "Network Usage" is more than 100%. Here I is the example of such a email:

[b]Target: VMachine1

Old Status: Green

New Status: Red

Current value:

Virtual Machine Network Usage - (Metric Network Usage (Average/Rate) = 196%)

Alarm: Virtual Machine Network Usage

(\[Yellow Metric Is Above 75%; Red Metric Is Above 90%])

Description:

Alarm Virtual Machine Network Usage on VMachine1 changed from green to red[/b][/i]

How this machine can use more that 100% of network? How does it calculated? Is something wrong in my configuration? What should I correct?

Also, sometimes I am receiving alerts that CPU/Memory usage is using more than 100%, too, and I think it's the same issue as the issue with this Network Alerts that I described above.

Thank you very much in advance.

Best regards

oleg

[mbradleyde]

I've seen the same thing. I see it a lot with Disk Usage. My latest alert says that my disk usage was 263%. I would also like to know how it does this calculation. Does anyone have any input?

[RParker]

most likely this is a shares problem. It is going over it's allocated shares. If nothing else is using the allocated resource, and IF the ESX server can give it more, then it is allowing this VM to go over, that's why the numbers are skew.

How much NIC are you giving the VM, are you at the lastest version of VM tools, and what is the VM doing to spike the network.. Must be some sort of file transfer going on...

[RParker]

If you give say 1000 disk share to a VM, and your resource pool is setup for 2000 from the ESX server, other VM's in that pool may have a higher priority of the slice.

The VM can allocate more than its share \*IF* this amount is available, and no other higher priorty machines NEED the disk usage at the moment, your VM will then be 200% of usage (2000 is 2* the allocated 1000 share).

Thus 2 * 1000 = 2000 (200%). 263% is some calculation based on shares and total shares from the resource pool and the ESX share for that vM.

I don't know what your setup is, perhaps you can give more detail about:

1) VM disk share

2) Resource Pool share for ALL VM's in the pool

[eric_heilig]

Actually, I am seeing similar alerts for my ESX servers on both network and disk I/O. Please see the alarm output below. Anyone have input on what this might be?

Regards,

\-- Eric

Target: eg-farmzp-a001

Old Status: Green

New Status: Red

Current value:

ESX disk IO - (Metric Disk Usage (Average/Rate) = 355%)

Alarm: ESX disk IO

(\[Yellow Metric Is Above 80%; Red Metric Is Above 90%])

Description:

Alarm ESX disk IO on eg-farmzp-a001 changed from green to red

[kpear]

I don't even have resource pools or share modified from their defaults, and my alarms note up to 1205% disk usage. How is this possible? Does this mean that a single VM uses up more than 12 times its internally alotted share (if it even works this way)? The ESX has 12 VMs, does this mean it borrowed all of the other VMs share somehow considering that the alarm reported 12x the 100% disk usage? The datastore browser indicates that the disk is only about 50% full. Please advise. VC has become the "server that cries wolf." I now tend to ignore the alarms altogether, which is not good.

[frank_wegner]

This is a bug in the displayed units of the VI Client.

Disk metric monitored by alarms is "kiloBytesPerSecond", not "percent".

Network metric monitored by alarms is "kiloBitsPerSecond", not "percent"

F

[eric_heilig]

Kpear,

I have followed a support thread with VMware and below are the responses I received. It seems that this is a bug in VirtualCenter and that the metrics reported for network and disk IO are absolute values and not percentages. This bug has not yet been fixed and I was unable to confirm when it would be addressed.

Regards,

Eric

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Hi Eric,

Looking further into these large percentages, it is apparently a bug in

VC at this time. The numbers for disk IO and network are actually

absolute (kb/s) and are not percentages. This will be addressed in an

upcoming release/patch of VC.

Regards,

VMware Technical Support

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Eric,

It will most likely be addressed in the next release of VC which may be

later this year. Unfortunately, I do not have a hard date, and I am

unable to guarantee that it will come out by year's end.

Your case number has been added to the bug report for this issue. I

will go ahead and close out this case unless you have any other

concerns.

Regards,

VMware Technical Support

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

[frank_wegner]

The screenshot in step 7 on page 309 of the Basic System Administration Guide for ESX Server 3.5, ESX Server 3i version 3.5 Virtual Center 2.5 (RC version: ) still shows network metrics in % in the alarm settings.

I see are two options (under the assumption that this does not change between RC and GA):

a) It is a bug in the documentation

b) This bug is not fixed in ESX 3.5 / VC 2.5

I have informed VMware engineering about this. No feedback as of today.

[OaktonIT]

It's been a while, any updates to this?