Network slowness arises after introducing additional uplinks to virtual connect

[MohdAlim]

Greetings

I am experiencing network slowness after introducing additional uplinks to the virtual connect.

Previous configuration:

Bay 1: Port X5: 10G --> SW1

Bay 2: Port X5: 10G --> SW2

No slowness detected with above.

New configuration:

Bay 1: Port X5: 10G --> SW1 (LACP Port Channel 13)

Bay 1: Port X6: 10G --> SW2 (LACP Port Channel 13)

Bay 2: Port X5: 10G --> SW1 (LACP Port Channel 14)

Bay 2: Port X6: 10G --> SW2 (LACP Port Channel 14)

Slowness detected with new uplinks. After removing uplinks, network went back to normal. FYI, the Cisco switches have been configured with VPC.

Has anyone encountered such issues. Any advise is greatly appreciated.

Thanks in advance.

[Hongjun Ma]

what's the OS on blade? can you isolate to see if po13 or po14 has the issue or both port channel has performance issue? For vmware, a given vm is always hashed to a given vmnic in normal condition using default vswitch hashing algorithm.

please take a look at "show int po13" and its physical links, do u see any obvious interface errors/drops? same thing to po14.

[MohdAlim]

The blades are running vSphere 5. It's happening on both the port channel. Let me elaborate how this unique problem was detected. We have several VMs (Redhat) running on the ESX hosts with. These VMs have a mapping to a NFS storage. W/out the additional links, when we execute the command df -h from the VMs, the output is instantaneous. With the additional links, it will take like 10 seconds or more, or even hand just to execute the df -h command. Once removed, it goes back to normal.

I will try to check if there are interface errors/drops.

[Hongjun Ma]

can u shut the link on X5 for both modules to see if you still see the problem?

I'm trying to see if this is the problem with individual link or only happens when LACP link bundle is in place.

btw, I'm assuming all 4 links are in active/forwarding state for VC(shared uplink set) view.