Nexus 1000v Split-Brain Resolution

[ngtransge]

Hello,

I am interesting in Nexus 1000v Split-Brain Resolution. In documentation they speak about "proper configuration" during Split brain resolution.

http://www.cisco.com/c/en/us/td/docs/switches/datacenter/nexus1000/sw/4_2_1_s_v_2_2_1/high_availability/b_Cisco_Nexus_1000V_High_Availability_and_Redundancy_Configuration_Guide_2_2_1/b_Cisco_Nexus_1000V_High_Availability_and_Redundancy_Configuration_...

Could you explain what they mean in proper configuration? In what cases could VSM get improper configuration?

When split brain resolution chooses VSM that should be rebooted, how long VM attached to its VEM lose network connection?

Thank you

[sprasath · Accepted solution]

Hello,

Proper configuration here just means the latest/correct configuration. There could be a chance where one VSM has the latest configuration (proper config here) and the other one has a different configuration (an older config maybe).

So the split-brain logic considers the various parameters mentioned and chooses an active VSM to continue. The other VSM reboots and joins back as standby.

Ideally, when the chosen standby is being rebooted, there should not be any connectivity issue to the VMs as they'll be attached/programmed by the active VSM.

Thanks,

Shankar

[ngtransge]

Thank you for explanation.

In case there is split brain between VSM.  Besides Nexus administrator, could it happen that one of the VSM configuration is changed/updated ?

[sprasath]

Split brain or not, any configuration change has to be done from the VSM. So anyone with access to make changes on the VSM can make the changes.

'show accounting log' has a good enough buffer and captures config changes.

Thanks,

Shankar

[ngtransge]

Hello sprasath,

Thank you for helping. I will try to clarify my previous question.

Let’s say there is split brain between VSMs. And on second site there is second VSM located, which was secondary in VSM HA. Also additional failure has happened on second site and one of the ESX host failed, and all VMs of that host are automatically started on other ESX host on that site. Let’s say there were VMware HA cluster environment.  So all VM have moved from one host to another automatically.

From VSM point of view, could this incident be considered as configuration change?

Because VM have moved from one host another.

Thank you,