Networking working only one one blade [UCSM + VMWare issue]

[Ron Baduach]

Hi ,

I'm a Cisco worker , and we got some issue .

I'll start to explain it from the VMware side , and then will move to the UCSM side .

We got DVS on VMware, and there is strange issue that some vlan is working only on 4-5 hosts , and not on the others ...

We did a little troubleshooting , and we saw that the hosts that succeeded to working with this Vlan, are from the same blade .

Notice, that all of our blades were created from the same Service-Profile Template , and they all got the same Vlans .

Hope i explained it well .

Someone have any idea about it ? what can i check ?

thanks in advance ,

Ron 

[Walter Dey]

Hi Ron

I can hardly believe that this is related to UCS, or SP. Are you sure that your ESXi installations are all identical ? Can you ping the ESXi that have a problem ? Is HA and loadbalancing identical on all ESXi hosts ?

Walter. 

[Ron Baduach]

Hi Walter, thanks for your response .

OK , got it .

I can ping the ESXis that have problem .

The hosts [the "good" ones , and the "not good" ones] ,are in the same cluster , they all got the same HA , but about the loadbalance , how can i check it on specific host ? I know only to check it on the whole Cluster , on the "Cluster Settings"

Thanks

[Walter Dey]

Are the ones that have problems using exactly the same North bound uplinks, like the one which are working ok ?

Can you vmotion a VM from a failing ESXi to a good ESXi, and then it works ?

Are both UCS fabrics operational ?

[Ron Baduach]

1. Same northbound uplinks

2. I can VMotion VMs on another Vlans from "working" to "non-working" ESXi [I can also VMotion on this Vlan , and it'll success , but the VM's networking will be down]

3. both fabrics are operational [one is primary , and the other is secondary]

[Walter Dey]

Reg. 2) do I understand you correctly, the problem moves with the VM ? if yes, I would compare the vnic configuration, eg. port-group,.....

[Ron Baduach]

Hi , thanks again for the assistant & time

The problem is not to move the VM, the VM is success with migrating between all the hosts, but, the VM's networking [ping] is working only on 3 hosts ....

[all hosts connected to same VDS]

[Walter Dey]

So a VM which cannot be pinged is moved to another ESXi host, and there is works ? correct ?

- which ESXi version

- which UCS version (I assume the proper vnic/fnic drivers are installed, same on all blades)

- are the blades that are working and not working in the same chassis; or are the non working in another chassis ?

[Ron Baduach]

See inline in Bold :

- which ESXi version - ESXi 5.5

- which UCS version (I assume the proper vnic/fnic drivers are installed, same on all blades) - UCSM 2.2(5a), and the FIs version : UCS-FI-6248UP . the same drivers installed proper, and it's the same on all blades

- are the blades that are working and not working in the same chassis; or are the non working in another chassis ? The working blades are from the same chassis, and all the non-working, is from other different chassis [but same model of chassis/blades]

[Walter Dey]

Did you install on local disk, or boot from SAN ?

Very strange ! Is it possible, to move one blade from a non working chassis to the one which seems to be ok ?

[Ron Baduach]

All is the same , local boot .

mmm.... Nice idea, I'll check with my manager