Lost vLan

[TBKing]

13 HP blade servers in a VM- ESXi5 cluster.  2 of the blades are in Blade Center #2.

All servers have:

2 paired/redundant nics for vMotion and vMgmt, on 2 vLans

2 paired/redundant nics for Data network supporting 3 vLANs

Without any apparent reason, One of the Data vLANs stopped working for the two blades in BC#2.

It is the "Default" vLAN ( "0" ) for that segment.

The other vLANs continue to work fine.

vLan 0 works fine in the other blade center.

vMotion, management, console, etc all work.

We've rebooted the ESX boxes and the HP blade switches and no change.

Nosed around logs, configs, etc and nothing apparent.

ESX does not report a disconnect or any other issues.  Infact, before I remembered to turn off DRS, it pushed one of the affected VMs back while I was clearing for a reboot :smileysilly:

Because it's both blades in one chassis, I'm leaning towards a chassis issue, or the next step out.

Network guys say "Not Us", VM guys say "Not Us"

.. any ideas where to look or pin this down further?

Thanks.

[iw123]

Hi,

When you say, stopped working, do you mean the virtual machines configured to use that portgroup can no longer access the network?

[TBKing]

Yes - They lost connectivity to the network - but the VMs do not show a port-down, disconnect or anything apparent like that.

[chriswahl]

Have your network team confirmed that VLAN 0 is the native VLAN for the trunk port going into your chassis #2. In Cisco IOS, this is basically issuing a "show interface gi#/# switchport" command.

Also, since you mentioned this is HP gear, make sure that you have the default VLAN value configured to match in Virtual Connect.

[iw123]

Hi,

Can those virtual machines communicate with others on the same subnet?

[TBKing]

The VMs on the same host on the "lost" vLan could talk to each other, but not to VMs on the other blade in the same chassis.

It turned out to be a config on a switch upstream of the blades was apparently not passing that subnet as vLan 0

Thanks for your replies.