Assumed AD configuration when using ADConnector

I am using ADConnector with midPoint to implement live provisioning.
Occasionally, discovery occurs and the following events occur.

• Shadow data in dead state is created and NO_OBJECT error occurs.
• The AD side detects the conflict and creates "CN=(username)\0ACNF:(objectGUID)" which is different from the original "CN=(username)".
• If this happens, the ENTRY_EXISTS error will continue in subsequent live provisioning.

The environmental conditions here are as follows.

• AD provisioning is set up so that midPoint user registration -> AD registration is enabled.
• When midPoint user is registered, SecondaryChange is added and updated in the hook script.
• As a result, AD registration → AD lookup → AD update is performed in one midPoint user registration.
• AD is a redundant configuration of DNS round robin.

We are still investigating, but there is a case that midPoint refers to both AD1 and AD2 in a single process.
I'm guessing that this is happening because there is a time lag in replication between AD units 1 and 2.

In the first place, does it support DNS round robin configuration?
Also, I'd like to know the expected AD configuration when using ADConnector.