Assume that you have database mirroring set up between a principal server and a mirror server, and that the principal and mirror servers are running on Microsoft SQL Server 2012, 2014 or 2016. During the redo phase of a large transaction on the mirror server, the redo thread cannot set an exclusive (X) lock and this generates an assert memory dump file. This issue may not be resolved even if you try to restart the SQL Server service on the mirror.
This issue is fixed in the following cumulative updates for SQL Server:
Microsoft has confirmed that this is a problem in the Microsoft products that are listed in the "Applies to" section.