switchover can freeze from 10.5 to 10.6 using event scheduler on wrong system table in 10.6

[svaroqui]

Using following replication-manager setting

# Failover activate and disable event scheduler
failover-event-scheduler = true
# Failover change the event status ENABLE OR DISABLE ON SLAVE
failover-event-status = true

Following a package upgrade in debian the mysql_upgrade script is not auto run this can lead to a candidate master with scheduler OFF at startup but still ON in config

doing a switchover in such case with active events on the old master , will put replication-manager in an infinite loop waiting for the replication stream to catch but the replication is broken because ALTER EVENT DISABLE ON SLAVE will failed inside replication

Workaround could be to check if elected master have event_scheduler on or if putting to on is failing to avoid switchover. need to investigate if it's not to late and what are the consequences to turn it on , also to be less intrusive on ALTERING EVENT with disable binlog but this may lead to DDL divergence from master and slave regarding event definition