Create a Post
cancel
Showing results for 
Search instead for 
Did you mean: 
Participant

VSLS/VSX in the loss of the sync connection

We are deploying a 3 member VSX environment. The 6 Virtual System instances (and a virtual switch to connect them) are distrivuted via VSLS. Two of the members are physically at one site, and the 3rd appliance is a little distance away in another DC. In the steady state, we have 1 VS running at that remote site, the other 5 VSes on the two appliances in the main DC. All this works well for the most part.

From what we see, if we disable or removed sync connectvity to an appliance, all of the VSes on that single appliance go to DOWN state, and the other two take up the load. Which we sort of expect. 

So if the interconnection between the two site fails, the two members that are together stay up and the remote one all by itself goes to down. Is there any way to change that behaviour? Also can we have a VS that always only runs on that remote site regardless of whether the sync network is up or not?

 

(The VSX admin guide seems to only talk about failure modes such as loss of members or the monitored interfaces, rather than the cluster interconnect. )

0 Kudos
Reply
1 Reply

The Sync interface is also monitored by ClusterXL. If CCP packets get lost here, a cluster error is triggered. This is normal behavior.
To avoid this, a bound interface can be used for the sync.

0 Kudos
Reply