ESX cluster loses network connectivity if change to layer3 switch

Experts, I have a connectivity issue with VMWare that I can't seem to resolve.

I have an ESX 5.5 lab that's currently working as configured under a "Router on a Stick" scenario.

ESX Host  *Vlan 2
VMs *Multiple Vlans             --> Layer 2 Switch --> Layer 3 Router (Gateway 10.x.x.x per each vlan)
iSCSI *Vlan 8
NAS *Vlan 2

Everything works fine, but as the L2 switch is capable of L3, I'd like to simplify the configuration, and get wire speed between the different subnets.  Unfortunately, when I remove the above configuration and set the gateway IP addresses up as a virtual interface on the switch.  This is where it breaks.

ESX Host  *Vlan 2
VMs *Multiple Vlans             --> Layer 3 Switch (Gateway 10.x.x.x per each vlan) --> Layer 3 Router (Sending default route)
iSCSI *Vlan 8
NAS *Vlan 2

Moving the L3 back from the router, and into the switch - works on ALL of the devices (iSCSI/NAS, etc.) - but the ESX hosts immediately drop connection and I can't reconnect to them at all.  With the updated configuration, I'm able to access the iSCSI devices, as well as the NAS boxes that I have, even in the same server network as the hosts/vm's I'm no longer able to communicate with.

All Vlans I have in the scenario, they're all being tagged down to the esx hosts on all ports.  The only configuration change in any of this, is me moving the gateway IP from the router, down to the L3 switch.

I'm decent in router configurations / as well as esx - so I'm not really needing any configuration guides, but I could use the "powers of the minds" here to reasons as why the hosts would stop talking the way they do, just by changing the gateway to a different device.  I'm leaning towards the issue being arp cache, but can't confirm this is the issue, as the other devices in the same vlans have no problems after the change - only ESX is exhibiting this behavior.  That, and I've power-cycled the hosts after changeover too.

Would anybody have any ideas on where I can point the "stick of troubleshoot" at?  *FYI, the L3 switch is a Brocade 648P-POE, but not that it matters.
LVL 5
usslindstromAsked:
Who is Participating?
 
usslindstromConnect With a Mentor Author Commented:
That was exactly it.  VMWare has articles describing the issue, for those interested:

Cmd introduced in ESX 5.5 (http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2049667)
0
 
usslindstromAuthor Commented:
I'm awesome.
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.