I am using the QAN-agent for my home database in the performance schema mode. Normally, the agent should do this query:
SELECT DIGEST, CURRENT_SCHEMA, SQL_TEXT FROM performance_schema.events_statements_history
at every second since the ticker for the worker is defined as:
w.ticker = time.NewTicker(time.Millisecond * 1000)
At about every hours, it seems that a new worker is instantiated and the rate increase by 1/s (see the graph). I confirmed additional threads using "strace -fp pid-of-qan-agent" After a few weeks, my home database was doing 300+/s such query.