ProLiant Servers (ML,DL,SL)
1748184 Members
3962 Online
108759 Solutions
New Discussion

Re: Virtual Switches suddenly stop functioning

 
SOLVED
Go to solution
ICS_CS
Occasional Advisor

Virtual Switches suddenly stop functioning

Several months ago we got a call from a client with a standalone 2012 R2 HyperV host: Absolutely NO network traffic was going through any of the virtual switches, yet we could see no errors or warnings of any kind in any of the logs.  Desperate to get them back up and running we simply deleted all the virtual switches, recreated them (identical to what they were) and reattached all the VMs.  Note that the port reserved for the host had no issue.  At first we chalked it up as a fluke....except that it happened again to a completely different customer.  And then third.   And then a forth.   And then several months after the first incident it happened againt to the first customer.   It has become a chronic issue which happens ever few months and no trigger can be found.

 

The only thing common to all of them is A) Server 2012 R2 host and B) HP ProLiant G8 servers (various models, inluding DL360, ML350, and ML310).

 

All hardware has the latest firmware and drivers as prescribed by HP.  All Windows updates are applied in timely fashion.  No event, change, or tampering precipitates the event.  The virtual switches just stop passing data.

 

I've reached out to the Microsoft community on technet but no one there has seen this behavior before.   I hope someone here can shed some light on it.

 
2 REPLIES 2
Arnaud Delaloy
Valued Contributor
Solution

Re: Virtual Switches suddenly stop functioning

hello,

 

We had a quite similar incident, on non-blade proliants : some hyperV VMs juste lose connectivity, with no message whatsoever.

 

 if it belonged to an hyperV cluster, moving the VM to another node made it resume its network traffic, but putting it back on the first node made the network stop again.

  after rebooting the offending node, or disabling an re-enabling the NIC dedicated to the VMs , network would be fine for a few days...

 

 we placed a support call on Microsoft; turns out there is an issue with VMQ fonctionnality and the Broadcom NetXtreme NIC Driver (happened to us on gen8 DL360 and ML350):


http://support.microsoft.com/kb/2986895/en-us

 

 we disabled VMQ on the node level (no reboot needed, only a short network disruption) and the issue went away !

 

ICS_CS
Occasional Advisor

Re: Virtual Switches suddenly stop functioning

Thanks for that.  We'll try it and report back.

 

Cheers.