I understand the change is not applied. But when running the script on the slave server, it should tell you wether it can connect or not and if it can, set the debug level. So please run the script on the slave, not the master, and view the output. You can set it manually but this will cause differences between master and slave in the database so it can break things. If you want to do this, open the database on the slave, go to the server table, and under config, change loglevel to =0. But again, this is risky and I'm afraid it won't help you much.
Well, indeed, changing the loglevel in the database did not help but you gave me a crucial hint mentionning this database! As it seems, there was a discrepancy in the "updated" column between the master and the slave, wtih the slave ahead. Putting it back in sync solved the issue. Thank you very much ... and happy new year.
For what it's worth, this loglevel is read directly from the master database, so even though the datalog queue is hung up, the log level change is still effective when you run server.sh on the slave.