cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
26292
Views
5
Helpful
12
Replies

VIF link state down

Amr Hafez
Level 1
Level 1

Hi

When I configured service profile to a blade then deploy ESXi, It works correctly for a time then I found the following error " VIF 716 link state is down" as in the attached and the server management IP is pingable also the ESXi IP is pingable during the error

I did the Same to another blade but no problem

 

2 Accepted Solutions

Accepted Solutions

Hi Amr

I see the following issues

- if global connection policy is 2, it means that min 2 links have to be up, otherwise the chassis discovery will fail on this fabric, and not bring up the remaining link. Please change global default down to 1 link !

- In your SP, you have selected the "hardware failover flag"; please remove that for both veth interfaces. Best practise is to let vswitch handle failover, not UCS !

- is the IP pool used for KVM in the same subnet / vlan, as the out of band management address of the FI, and VIP

Walter.

View solution in original post

Hi Amr

This has been discussed many times before

see eg. https://communities.vmware.com/message/2183957

But about the HW failover configuration I really do not know why not recommended to use the HW failover I think it's better than  failover of VMware VSS

Agreed that it should work, but I'd prefer to enable and configure failover behavior in ESXi rather than UCS. In my research on this issue, I found a thread over on the Cisco UCS forum that has a clear statement regarding this configuration: https://supportforums.cisco.com/thread/2187653.

 

"Never use fabric failover with a hypervisor vswitch. Let the failover mechanism of the vSwitch handle the network outage."

 

So while it may be a valid design, it looks like the consensus is to let your vSwitch handle the failover.

Cheers

Walter.

View solution in original post

12 Replies 12

Amr Hafez
Level 1
Level 1

Additional info.

When i remove the IOM-1 cables to FI then connected them back, the VIF came up again

Hi Amr

Do you have a SP with 2 vnic's, one attached to fabric A, the other to B ?

I assume you also see on the vswitch, that one link is down.

How many links do you have between IOM and FI ?

Do both FI have North bound uplinks ?

It could be, that either the link between IOM and FI or all uplinks of that fabric FI are down.

Hi Walter

I hope you are fine 

you are really helpful

1- Do you have a SP with 2 vnic's, one attached to fabric A, the other to B ?

yes 

2- How many links do you have between IOM and FI ?

2 links per 1 IOM

3- Do both FI have North bound uplinks ?

yes, uplinks in the same core

I tried to unplug the FI-IOM cables and it work correctly 

but it's a big problem if we lose one link from FI to IOM some VMs and ESXi will be down , I mean no failover there? 

 

Saurabh Kothari
Cisco Employee
Cisco Employee

Hello Amr,

is the interface up on the OS side? 

do you lose services when the error come's up?

please check the following bug:-

https://tools.cisco.com/bugsearch/bug/CSCtt38889/?reffering_site=dumpcr

 

Thanks,

Saurabh

Hello Saurabh

the interfaces in ESXi side was up and can connect to it through vKVM and no services loss but the ESXi not pingable

no OS in the hyper-visor ESXi till now so i do not know the connection states

I supposed it's a bug may be

thanks 

Hi Amr

Does your vswitch have 2 links, one to fabric A, the other to B ?

How is you loadbalancing on ESXi done ? I hope it's "port id"

Which UCS version ? which ESXi version ?

Are both IOM links to FI up; if you not, you might have to ack the chassis (default gloabal connection is 1).

Walter.

Hi Walter

Does your vswitch have 2 links, one to fabric A, the other to B ?

yes as in attached img1,img2

How is you loadbalancing on ESXi done ? I hope it's "port id"

yes it's port ID by default I think see LB attached image

Which UCS version ? which ESXi version ?

6.0.0

Are both IOM links to FI up; if you not, you might have to ack the chassis (default gloabal connection is 1).

yes  both up and global chassis connection is 2. see attached global_discovery

Hi Amr

I see the following issues

- if global connection policy is 2, it means that min 2 links have to be up, otherwise the chassis discovery will fail on this fabric, and not bring up the remaining link. Please change global default down to 1 link !

- In your SP, you have selected the "hardware failover flag"; please remove that for both veth interfaces. Best practise is to let vswitch handle failover, not UCS !

- is the IP pool used for KVM in the same subnet / vlan, as the out of band management address of the FI, and VIP

Walter.

Hi Walter

thanks for helping

I modified the global discovery policy to be 1 link

But about the HW failover configuration I really do not know why not recommended to use the HW failover I think it's better than  failover of VMware VSS

Also, if we do not use ESXi and use OVM or redhat or HyperV  or use server 2012 baremetal installation, could we do the SW failover in those cases?

yes the KVM IP and Vlan is the same in FI mgmt

Amr 

Hi Amr

This has been discussed many times before

see eg. https://communities.vmware.com/message/2183957

But about the HW failover configuration I really do not know why not recommended to use the HW failover I think it's better than  failover of VMware VSS

Agreed that it should work, but I'd prefer to enable and configure failover behavior in ESXi rather than UCS. In my research on this issue, I found a thread over on the Cisco UCS forum that has a clear statement regarding this configuration: https://supportforums.cisco.com/thread/2187653.

 

"Never use fabric failover with a hypervisor vswitch. Let the failover mechanism of the vSwitch handle the network outage."

 

So while it may be a valid design, it looks like the consensus is to let your vSwitch handle the failover.

Cheers

Walter.

the interfaces in ESXi side was up and can connect to it through vKVM and no services loss but the ESXi not pingable

How is you management vlan for ESXi setup:

2 choices

- either you define vlan id on ESXi, and mark it in UCS as non native

- you don't specify any vlan in the ESXi configuration, and mark it in UCS as native vlan

Hi Walter

didn't specify any vlan in the ESXi configuration, and mark it in UCS as native vlan

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: