Connectivity randomly breaks between UCS and 3560 Switch

[syed.mohsin]

Hi

We have established connectivity between UCS Servers and Cisco 3560 switch. Two Trunk ports are configured and required VLAN is allowed over trunk. everything is up and running. Virtual machines are reachable. 

Switch port configuration:

interface GigabitEthernet0/27
 description **** Connected with FIA-B ****
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan x
 switchport mode trunk
!         
interface GigabitEthernet0/28
 description **** Connected with FIA-A ****
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan x
 switchport mode trunk
!        

but sometimes randomly readability breaks from switch to UCS. After accessing VM from console connectivity itself get restored.

Has anyone experience this issue? what could be the problem? any solution to this?

Thanks

[Kirk J]

Greetings.

Please check the mac aging for the guestVM macs in question on the 3560.

Is the ESXi host mgmt IP still reachable when the guests are no longer pingable?

I would start with general stuff ,VMware tools up to date in guestVMs OS, ESXi host nic drivers up to date (typically igb or bnx2 if rack servers).  Switch ports set to portfast so spanning tree events don't flap ports.

If linux OS in guestVMs, run tcpdump to see if you are getting some of the traffic (icmp from switch), or wireshark if windows OS.

Thanks,

Kirk...