We have a DB in stress environment:
SQL Server 2014 (SP1-CU3) + Always ON
The whole db is around 850GB, but in some point something happens and in memory FG is growing to 250GB and a log to 450GB (original sizes around 1 and 5GB).
It happens fast - I can see it by checking log backups in 5 minutes backup files are growing from 20kb to 200mb and in another 5 to 500MB and "stays" in this area. Log stays full and cannot be shrunk.
Log reuse wait shows
After removing this DB from AON log reuse wait changes to
XTP_CHECKPOINT, but still nothing happens.
Even changing recovery model to simple doesn't help.
Only after executing checkpoint, in memory FG size reduces slowly till around 1GB size.
Afterwards the log can be shrunk also.
Does anyone familiar with this behavior? What would you suggest to monitor to find a root of this problem?