In Microsoft SQL Server 2016 and 2017 Always On Availability Group (AAG), when the data seeding feature is enabled, the following issues occur:
- HADR (High availability and Disaster recovery) worker threads suddenly increase a lot and reach exhaustion with an error message:
"The thread pool for Always On Availability Groups was unable to start a new worker thread because there are not enough available worker threads."
- CMEMTHREAD wait type occurs.
- Massive transaction blocking.
This issue is fixed in the following cumulative updates and service pack for SQL Server:
Service pack information for SQL Server 2016
This issue is fixed in the following service pack for SQL Server:
Microsoft has confirmed that this is a problem in the Microsoft products that are listed in the "Applies to" section.