5596UP killing Vcenter "heartbeat" packets on Lan and WAN

[dmooreami]

Just moved my entire off a 6509 to the dual 5596up's running version 6.0(2)N2(5) with UCS.

Now, Vcenter5 (physical machine not vm) is having problems with its "heartbeat" to tcp/udp 902.

It isn't a routing issue, not a "arp issue", not a issue with  a lower switch having " ip device tracking " enabled. Not an access-list block, not a firewall block. 

Every minute we can watch Vcenter lose contact with local VM's that are on the UCS  which is connected by Four 10-Gig trunks.

All other protocols and devices have no issues. Its just Vcenter on the local lan and going out across our WAN.

Had zero issues with this when I had all my core routing running on a 6509 running 12.2sx code. 

Nothing changed on the Vcenter side or on our UCS.  The routing was just moved to the Nexus 5596UP's.

[dmooreami]

More information. 

My Vcenter is on the same subnet as my esxi servers that run the VM's.

No routing is taking place.

[Walter Dey]

are you talking about "VM Monitoring (aka VM HA) heartbeat" ?

[dmooreami]

VM monitoring are the TCP/UDP probes that Vcenter sends out to the VM servers ever 20 seconds.

This lets vcenter know that the VM servers are working correctly. Without a VM heartbeat, Vcenter thinks the Esx server is down, so you can't manage any of the VM hosts on that ESX server or migrate over VM's from one ESX server to another.

[Walter Dey]

Any information in the /var/log/vmware/hostd.log file on the ESX host to see if it was caused by VMM.

[dmooreami]

Problem was in Vcenter. Required a rebuild of Vcenter onto the UCS blades.  We had Vcenter running on a standalone server, not as a VM on the UCS.