ignore_external_master is ON, and a failed master with old external connection tries rejoin, which causes master to be lost

2/5 19:36:05 on app-1 (10.16.1.18): systemctl stop mariadb

App-0 (10.16.1.17) is promoted to master
Fixed replication 10.16.1.10 <-> 10.16.1.17
2/5 19:37:40 on app-1 (10.16.1.18): systemctl start mariadb

App-1 is re-joined to cluster as Slave
Success

2/5 19:42:38 on app-0 (10.16.1.17): systemctl stop mariadb

App-1 (10.16.1.18) is promoted to master
Fixed replication 10.16.1.10 <-> 10.16.1.18
2/5 19:44:38 on app-0 (10.16.1.17): systemctl start mariadb

Maxscale loses track of 10.16.1.18 as master
FAILURE – maxscale has no master

2/5 19:48:37 on app-0: echo “STOP SLAVE;” | mysql –-password=*****

Maxscale fixes itself and 10.16.1.18 is again master
App-0 is “Running”
2/5 19:49:27: on app-0: echo “START SLAVE;” | mysql –-password=*****

App-0 is “Slave, Running”