Idle Client receives an error mysql has gone away after 10 minutes

We are using a script that create an index on a large table. The script ends with an error saying "MySQL has gone away" after exactly 10 minutes.
From what we see in the network packet capture we made, the Maxscale server is sending a TCP FIN/ACK after 10 minutes of inactivity of the client.
The query to create the index is still running on the cluster after Maxscale has closed the connection with the client and ends correctly.
If we change the parameters of the script ot execute the query directly on the cluster without using Maxscale, everything is working properly.