How to manage special 2 node configuration

[Peter Mueller]

> I don't think nice_failback is relevant.
> 
> Immune to *which* situation?
> 
> Cluster partitioning?
> 
> The bug in cluster rejoining?
> 
> No one has yet reported testing the bug fix to the cluster 
> rejoin problem.
> 
> I suspect it works, but don't have any way to be sure...
> 

cluster rejoin bug, I believe.

If I don't have a dual LVS QA setup to test mon with serial+eth0
heartbeat... what is the recommendation?  At this point I cannot afford to
whack production.

thanks guys..