Virtual Connect Multi-Enclosure Stacking - What happens to vc domain when primary enclosure is down?

[chuckk281]

Glenn had a customer testing for Virtual Connect MES failover and saw some behavior they didn't understand:

**************************************************

Looking for information on Virtual Connect Domain behaviour when configured for Multi-Enclosure Stacking, and the base enclosure is not available.  The Multi-Enclosure Stacking reference guide provides some examples and the customer has implemented ME Mesh Stack 2 Enclosure: which “This configuration provides the ability to sustain a chassis, module, uplink or upstream switch failure/outage and still maintain connectivity to the network. If additional bandwidth is required, uplinks could be added to these VC networks, or additional VC networks could be configured.”

As part of their DR/Availability testing, the customer removed all power to the base enclosure to see if the environment could sustain an enclosure failure.  They are reporting to me that whilst the base enclosure was off;

-           servers in enc1 lost network connectivity via enc1 uplink ports.

-          Flex-10 Modules in enc1 did not take over VCM functionality, VC Domain was unavailable

Can someone please confirm what is the expected behaviour of the remaining enclosure(s)

*********************************************************

Carl explained what should be the likely behavior:

*********************************************************

Base enclosure “unavailable” is a very broad term. In the case I think you are describing – the entire enclosure is unavailable – as in power failure or disconnected from everything. This type of situation should not happen to a single enclosure, since a total power failure is the only thing that could take a properly designed infrastructure down. In that case a single enclosure is not going to be the only victim of a power failure – all enclosures would be affected.

Given the scenario you outlined - as should be expected:

1. All remaining chassis will continue to operate in a degraded state on the remaining uplinks – UNLESS the uplink design is flawed. (See #3)
2. The master domain has control of all administrative functions. While the master is down, the VC domain will not be able to be administered.
3. Any uplinks on the master domain will be unavailable, which is why multiple uplinks are created on different enclosures, in order to avoid a single point of failure.
4. When the master is brought back online, normal administration will be available.

*************************************************

Does this impact how you have done MES for failover? Can you share best practices?