Hypershift network validation fails when scaling up the nodepool

When using RHCAM 2.9.1 on OCP 4.14.9 to create a hypershift cluster with 4.14.9. 
While scaling the  nodepool's replica from 1 to 2, the network validation of assisted service fails. 
This blocks the 2nd node join in the hosted cluster. 
However, the same network config works fine with agent-based installer.    

4.14.9    

N/A    

The following steps can reproduce this issue:
1. Create HCP
2. Config nmstate config with 2 bonding for 2 nodes
3. Scale node pool to 1
4. wait for several hours
5. Scale node pool to 2
6. Network validation fails

Second Node unable to join the cluster    

Node should be able to join the cluster