Stephane VAROQUI

Hi, when you say failover you mean switchover correct , you do not kill the master replication manager is doing exactly what your saying , FLUSH TABLES WITH READ LOCK...

Here is the code in 2.0 that do this. if cluster.master.DBVersion.IsMariaDB() && cluster.master.DBVersion.Major > 10 && cluster.master.DBVersion.Minor >= 1 { go func() { var err2 error err2 = dbhelper.MariaDBFlushTablesNoLogTimeout(cluster.master.Conn, strconv.FormatInt(cluster.conf.SwitchWaitTrx+2,...

so you have log_slave_updates in the servers ?

Can you also extract the binlog of the old master and tell us what is in position GTID 0-1-878021

I'm sorry i'm not able to reproduce so far under sysbench!

Hi, I would test using a proxysql along side replication-manager . Also can you provide the config file of repman? proxysql follow read-only flag. Do you see read-only set to...

a certificate won't solve the security risk , it is the same role as a password but adding on top a trusted authority and an expiration date. I will have...

May be something like a 2 Step Authentification to retrieve those password by SSL

Hi Elliot, Many thanks for founding this, we most of the time work on systemd, so the contribution of your script is very welcome ! Tx /svar

There is also a replication-manager-cli status command that could help to get the state of the server init script!