ESX Ping failed after 5 minutes on N5548/N2248

[mattofi69]

Hi Guys,

I have an issue regarding ESX servers which are connected on Nexus 5548 running version 5.0(3)N1(1a) and their 2248 FEX modules.

We have several ESX deployed on C4948 using trunk interfaces, without Etherchannnel.

I’d like to deploy new ESX on Nexus 5K infrastructure.

To do that, we started with the simplest configuration as possible: Only one interface connected to one N2248 with the truck configuration below :

interface Ethernet101/1/6
description Server Admin
switchport mode trunk
switchport trunk native vlan 152
switchport trunk allowed vlan 81,152
spanning-tree port type edge trunk
logging event port link-status
logging event port trunk-status

Below, the architechture deployed :

->The server is available on the network for about 5 minutes then stops responding.

Actions / Tests already done:

• •- one of the server has been installed with Windows 2008 R2 -> The issue still occurs.
  
• •- one network interface of the server has been connected to another network switch (Cisco C4948) and the others are connected on Nexus -> The issue still occurs.
  
• •- disconnect / shut down all network switch ports except one -> The issue still occurs
  
• •- configure the interface in auto neg mode (instead of force in 1000/Full) -> The issue still occurs
  
• •- configure the interface with a port -> The issue still occurs
  
• •- remove the flow control on the interface with the flowcontrol send off command -> The issue still occurs
  
• •- remove the native vlan command -> The issue still occurs
  

• •- configure the switch port in access vlan (no trunk) -> It works
  
• •- move the server to another rack to connect all interfaces to cisco C4948 -> It works
  

Any idea on this problem ?

Is this config, with trunck interface, be correct ?

The goal may will be to have a vPC between the server an C2248, but I'd like validate step by step each point, before deploying vpc config in production environment.

Thanks for your help.

Cheers

[Steve Fuller]

Hi,

You show two VLANs on the trunk port configuration, but which of thees do you use when you configure the port as an access port?

For the VLAN you don't use on the access port, what is this for?

Regards

Sent from Cisco Technical Support iPhone App

[mattofi69]

Hi Steve,

Thanks for the reply. I use the vlan 152 when I configure an interface in access vlan, this is just for testing because in trunck mode, it's not working. On ESX side, the vmnic0 is configure with an ip address in vlan 152

Regards

[Steve Fuller · Accepted solution]

Hi Matthieu,

Here's an example of the configuration I have for an ESX server in my lab. The ESX server is configured via two NICs to a pair of Nexus 5548 on Ethernet 1/9. On the ESX server I have the vSwitch configured to Route based on the originating virtual port ID and a number of port groups defined, each with its own VLAN ID. A screen shot of the setup can be seen below.

On my switches the interface configuration is as follows:

ocs5548-1# show run int eth1/9

#

interface Ethernet1/9

  switchport mode trunk

  switchport trunk allowed vlan 171-178

  spanning-tree port type edge trunk

  spanning-tree bpduguard enable

  speed 1000

  load-interval counter 3 30

#

ocs5548-2# sh run int eth1/9

#

interface Ethernet1/9

  switchport mode trunk

  switchport trunk allowed vlan 171-178

  spanning-tree port type edge trunk

  spanning-tree bpduguard enable

  speed 1000

  load-interval counter 3 30

#

As far as the Nexus switch interface configuration is concerned, what you have is not very different from the configuration I have.

With the configuration you pasted above, the vSwitch port group associated with hosts in VLAN 81 would need to be configured with an appropriate VLAN ID. I presume that's the case, but please confirm.

The other point you make is that VLAN 152 is your access VLAN where your vmnic0 IP address is assigned from. As you have VLAN 152 configured as the native VLAN this will not be tagged on the physical switch and so will have no VLAN ID associated with it in the vSwitch. Do you also lose ping connectivity to the ESX management IP address in VLAN 152 or is it only the VMs IP address in VLAN 81 that are being dropped?

Can you also explain where the ping is between? Obviously a VM on the ESX host, but what is the other IP and where is that located?

Additionally, when you lose the ping response, how long does it take to recover or does it never recover?

Regards

[mattofi69]

Hi Steve,

Thanks for your help, I have followed your advice and I think we have resolved our issue !!

It seems that the solution consists to remove the native vlan command on Nexus interface and keep in place on ESX side, or remove the vlan tagged on ESX interface and keep in place the native vlan on Nexus interface.

Regards

Matthieu