Out-of-band management and HA A-P problem

Hello,

I experienced this issue with HA A-P cluster and would like to figure out what  exactly I did wrong. 

1. I have 2 FG-100F in A-P cluster with out-of-band management via dedicated mgmt interfaces.

2. I  tried to update them via Fabric management. The outcome is only secondary was updated and cluster becomes out of sync.

3. I removed secondary from cluster and updated primary manually. The cluster still out of sync.

4. I did factory reset for secondary and tried to join secondary to cluster. So I had clear config on secondary.

I put config of ha config block from primary, changing priority value. Set up mgmt interface as dedicated-to-management.

When I issued the commands:

 config ha-mgmt-interfaces

 set   interface "mgmt" 

an error occurred:  node_check_object fail!

I examined both config on primary and secondary and noticed differences : secondary has vdom "root" on mgmt interface, but  primary doesn't. So I backed up config from primary and restore it on secondary. After reboot devices built the cluster but with out of sync error. So I check ha checksum, found differences in application.list, corrected them.

The questions are:

1. Why  did node_check_object fail! error occur and how I can avoid it in the future? 

2. Why after clear update and restoring the same config did differences occur between config? Is it device unit specific problem?