Standby nic carries traffic?

[fletch00]

Hi,

We're transitioning from 4x1gb nic teaming to a 1x10gb active + 1gb standby config.

We really pleased with the performance BTW, especially when we upgraded central storage to 10gb.

But what I was surprised to see was the 1Gb "standby" nic reporting significant Tx/Rx 8mb/1mb traffic via vCenter and SNMP (cacti) graphs

Can someone please explain this (why a standby nic shows traffic) and point to the relevant VMware doc.

My expectation was it would remain idle until used.

One implication I think I encountered yesterday was I brought a misconfigured NIC into "standby" and it caused the ESXi 4.1 U1 box to lose connectivity.

thanks for any clarification.

[weinstein5]

Not sure - but first thing to check is configuration - because the port/port gourps will over vswitch settings - 

[WiscBrad2011101]

Can some kind of packet sniffer be used to identify the traffic ?

[JohnADCO]

You don't want it to MPIO?    That would seem odd to me, but when you look at your paths on your LUNs themselves?   Does it say round robin?  That is the setting that really makes it MPIO.    You have to check each lun too, under configuration, storage.

[AureusStone]

The standby NIC should not carry VM traffic.

Maybe it is just background noise?

Also make sure you have not set any of your port groups to override your teaming settings.

Also have a look at your switch.  You should be able to track down what nic the guests are being mapped to.

[JohnADCO]

I have to admit,  I assumed the 10g was the connection to his storage.   Probably a bad assumption on my part.

[fletch00]

This turned out to be a misconfiguration.

While the 1Gb nic was designated standby for the vSwitch there was a port group over-riding this which had the 10gb AND 1Gb designated as active.

This was immediately obvious when I did a cmd line

esxcfg-vswitch -l

which explicitly lists the pNics per port group (although it does not seem to distinguish active vs standby)

whereas the VI Client graphic which only shows the top level vswitch config (hiding any over-rides at port group level)

lesson learned - don't assume your port group traffic is following the top level vswitch config shown in the VI client.

Once I demoted the 1gb for that port group to standby, the 1gb traffic dropped to zero.

thanks

http://vmadmin.info