SQL Server 2016 SP2 Cumulative Update 12 contains some useful fixes:
SQL Server can shut down when you hit the max number of sessions
SQL Server can shut down due to lock conflicts during error message processing
“SQL Server crashes frequently” when you check a clustered columnstore index for corruption
AGs may have “interruption” – I guess that means the AG stops replicating, but it’s not clear from the KB article.
Stack dumps when transaction replication has a heavy workload on the publication database
Stack dumps when you query persisted computed columns
Stack dumps when you run a batch mode query with multiple joins (that’s columnstore indexes in 2016)
Scalar functions run slower than they did on SQL Server 2008 R2
Non-yielding scheduler when the primary AG replica runs low on memory
AG may think there’s a missing log block when the database isn’t very active
AG automatic seeding may fail
AGs with persistent log buffers: “all of the secondaries in the AG become unavailable”
Change tracking auto cleanup causes access violations and stack dumps
Access violations when Extended Events tries to capture query text on busy servers
Error when stored proc in database A pulls data from database B while being audited in database C
Stack dumps when you alter database-scoped configurations
Incorrect statistics histograms when they’re updated in parallel – which also means that after you apply this CU, you should probably update your statistics.
There’s also a new feature: the default system health Extended Events session can store way more data now, AND you can edit how much it holds!
