MySQL Router Failover - session on previous router existed in DB

[Nordiana Mohd Yusof]

Would like to seek your input on the following behavior observed in our MySQL environment.

Our setup is as follows:
-MySQL Router configured with Keepalived (VIP) for high availability
-Database connection timeout configured to 10 minutes
-MySQL InnoDB Cluster as the backend
-Max connection for 1 router - 900, max connection for DB 1000

During a router failover, we observed that the existing SLEEP sessions on the database remain active, even though the router has already failed over to the secondary node. Which contribute to the max db connection.

Is this the expected behaviour for MySQL Router and InnoDB Cluster? Specifically:
-Does a MySQL Router failover inherently leave existing SLEEP sessions on the database intact until wait_timeout/interactive_timeout is reached?
-Is there any recommended configuration or best practice to automatically clean up or minimise these SLEEP connections during or after router failover?

Your clarification will help us validate our understanding and tune our configuration accordingly.