After testing edge case RHCLOUD-41302, it was found that if a connector is created ahead of time, even when stopped, new hosts added before the migration are immediately replicated when the connector is runing. This can lead to hosts being removed or updated that dont exist in inventory yet, and then created during migration, causing inconsistency and potentially dangerous access approvals to resources.

The goal of this test is to see if using a blocking initial snapshot solves this issue but we need to understand what is actually blocked during the snapshot

Questions to Address:

• When the connector is created in STOPPED state and uses the default snapshot, what happens when a new hosts is added?
• If a new host was added while in STOPPED state, what happens when the connector is started? Is that new host added before the snapshot is done?
• Does a blocking snapshot ensure that the snapshot is done before all our changes (WAL changes like above)?
• Does a blocking snapshot prevent all writes to the database, or merely prevent the connector from syncing those changes until the snapshot is done and switches to streaming?