VM NICs lose connectivity after reboots

[CTSTech]

We have about 10 VMs running on an ESXi 5.0.0 (623860) host.  Whenever we reboot a VM, it will come back up but we cannot ping it from anywhere on the network.  We check the Windows TCP/IP settings, DNS and Gateway and they are all correct.  It cannot even ping itself.

If we disable the NIC in Windows and re-enable it - connectivity returns.

Any ideas?

[john23]

can you check the VMware.log for that vm where you are observing the issue?

- While rebooting??

- When you disable?

-When you enable?

-A

[a_p_]

Just to rule this out, please check whether "Connect at power on" is enabled in the VM's network adapter settings.

Other than this - if you are using a managed physical switch - take a look at the physical switch port settings to ensure port security isn't enabled and spanning-tree is is set to portfast (this may be different depending on the switch vendor).

André

[CTSTech]

a.p. wrote:

Just to rule this out, please check whether "Connect at power on" is enabled in the VM's network adapter settings.

Other than this - if you are using a managed physical switch - take a look at the physical switch port settings to ensure port security isn't enabled and spanning-tree is is set to portfast (this may be different depending on the switch vendor).

André

"Connect at power on" is enabled on all VMs.

I checked our Cisco Switch - port security is not enabled.  We do have spanning-tree enabled, but it is set to mode pvst.  Will turning on portfast effect anything else?

[admin]

Hello,

I would start looking at the logs and see if anything points out.. by any change you running on a Dv switch..?

-- Avinash

[a_p_]

In case of "mode access" and/or "mode trunk" ports for the ESXi host's VM-Network uplinks, spanning-tree portfast [trunk] will allow the moved MAC address to register faster on the physical port. The default spanning-tree algorithm can take up to ~45sec and may cause the issue you see. Setting portfast on an edge port should not cause any issues (I do this all the time).

see e.g. VMware KB: Sample configuration of virtual switch VLAN tagging (VST Mode)

André

[VirtuallyMikeB]

When a Windows box can't ping itself, it's likely a TCP/IP software stack problem - corruption of the OS.  Were all these VMs created from the same template?  If this is the case, can you build a fresh VM from installation media?  Even try a Linux variant and see if the issue persists.  If the issue is with the template, you'll want to rebuild the template from scratch.

All the best,

Mike

-----------------------------------------

Please consider marking this answer "correct" or "helpful" if you found it useful.

Mike Brown

VMware, Cisco Data Center, and NetApp dude

Consulting Engineer

[e-mail]

Twitter: @VirtuallyMikeB

Blog: http://VirtuallyMikeBrown.com

LinkedIn: http://LinkedIn.com/in/michaelbbrown