Port Group NIC Failover

[ronald_higgins]

Greetings Ladies & Gents,

I've got a a scenario where by a particular portgroup isnt failing over to the standby NIC, and i'm hoping the experienced VMware Admins

might be able to shed some light on the issue.

The ESX Host has 2 NICs, the NIC's are plugged into seperate switches (sw1 & sw2).

The ESX Networking Tab looks a little something like this:

#####################

ESX Config:

vSwitch0----

---

| Physical Adapters::| vmnic0 1000 Full | vmnic1 standby |

Virtual Machine Port Group

VM Network

3 Virtual Machines | VLAN *

VM1

VM2

VM3

Virtual Machine Port Group

monitoring

1 Virtual Machines | VLAN 60

VM4

Service Console Port

Service Console

vswif0: 196.x.x.x

#####################

NIC Teaming

The vSwitch , VM Network and the Monitoring PortGroup have the same NIC Teaming Settings.

Load Balancing: Use explicit failover order

Network Failover Detection: Link Status Only

Notify Switches: Yes

Failback : Yes

Active Adapters:

vmnic0

Standby Adapters:

vmnic1

#####################

Switch Configurations

interface GigabitEthernet0/7

description VMware esx01

switchport trunk encapsulation dot1q

switchport trunk native vlan 26

switchport mode trunk

load-interval 30

spanning-tree portfast trunk

interface GigabitEthernet0/8

description VMware esx02

switchport trunk encapsulation dot1q

switchport trunk native vlan 26

switchport mode trunk

load-interval 30

spanning-tree portfast trunk

All VLANs are allowed.

Everything works splendidly on the primary interface, and when we yank the cable out from eth0/vmnic0 everything (vSwitch0,VM Netwok, Service Console) switches over to vmnic1 barring the Monitoring Port Group which

for some reason looses connectivity when the switch occurs.

Anyone have any pointers on where to start looking why the Monitoring Port Group cannot shift across to the standby adapter?

Regards

Ronald

[ronald_higgins]

bump

[RenaudL]

What do you call "monitoring portgroup"? Are you sure its teaming settings are correct? Is there anything particular you see appearing in the logs? What does esxtop tell you?

[Rumple]

spanning-tree portfast trunk

load-interval 30

I am not familure with the size option on that spanning tree line and I am not sure what the load-interval is...

On the switches we have we are not using the size or load interval options and I don't even think we are using the dot1q encapsualtion line..but I could be wrong about that one...

no problems with failover/failback here.

[ronald_higgins]

Hi All,

Thanks for the response to date.

RenaudL,The Monitoring Port Group is just one that I have created for vlan tagging, this is the first VM that i have created that requires a non native VLAN.

If you're having problems with the description above, I've attached a screenshot of the settings.

Rumple, please ignore the , that came about when I edited the post, with the load interval, when you do a show interfac, you get a 30 sec instead of 5 min interval.

So after yanking the cable there wasnt any new log lines in hostd.log, esxcfg-firewall.log or esxcfg-linuxnet.log, esxtop didnt provide much help.

Basically, the problem is that I cant figure out why when

testing failover, anything that isnt in the native esx VLAN looses

connectivity.

The Networking Chaps tell me that once running on the secondary interface, the VM mac addresses are not being advertised, any ideas moving forward ?

[Rubeck]

Well, to me it sounds like the VLAN 60 dosn't exist on both switches..

Have you tried to configure a switch port as an access port, assign it to VLAN 60 and connect a labtop....?

Does this work?

/Rubeck

[Rubeck]

Ohhh... You could start by looking at network hints for both adapters.

/Rubeck

[ronald_higgins]

Thanks all,

Issue has been resolved, ended up being an incorrect port that the trunk was set up on.