Announcing SQL Server 2019 CU7, and Work Continues on Contained Availability Groups

Update 2020/09/24 – Microsoft removed SQL Server 2019 CU7 due to a problem affecting snapshots and CHECKDB.

Yesterday, Microsoft dropped Cumulative Update 7 for SQL Server 2019.

Among the fixes, Microsoft slipped in a new message in sys.messages. Message 47149 says:

Cannot start the job “%s” because this is secondary replica of contained availability group. Start the job in the primary replica of this contained availability group.

So, what are “Contained Availability Groups”, and why aren’t the C, A, and G capitalized? Those are both good questions. It doesn’t refer to using contained databases with Always On availability groups as it exists today since those don’t have Agent job support, at least not yet. It’s about AGs in Big Data Clusters, which can also contain system databases. It’s great to see work continuing on those because we’d love to have support for that in the regular SQL Server product.

Other fixes include:

• FIX: Concurrent inserts against tables with columnstore indexes may cause queries to hang in SQL Server 2016 and 2019
• When you run a SELECT query in Read Committed Snapshot Isolation (RCSI) on Clustered Columnstore Index (CCI) on a table in SQL Server 2019, the query may return incorrect number of rows under rare conditions
• Fixes for scalar UDF inlining issues (functions with CHECKSUM or OPTION RECOMPILE now work)
• FIX: Inconsistency occurs when ghost rows are inserted into mapping index rowset (“inconsistency occurs” is a polite way of saying corruption, I’m guessing)
• FIX: Intermittent Availability Group failover occurs as AG helper connection times out while connecting to SQL Server
• FIX: Distributed transactions may experience long waits with DTC_STATE wait type in SQL Server 2016 and 2019

And a lot of other good stuff. Go get ’em!