We have three situations that turn on InnoDB monitor output into the error log automatically: long semaphore waits, large amount of lock heaps, and long wsrep BF waits

While the former two print something like " Starting the InnoDB Monitor to print diagnostics." into the error log first, long BF waits don't do this, they just enable the output silently.

Also while the other two occasions take care of disabling the output at a later point in time, BF long waits do not do so. The log output gets enabled, but can only be disabled again by explicit {{SET GLOBAL innodb_status_output; }} or a node restart, so flooding the log with InnoDB status output every 20 seconds.