I have an XtraDB cluster (4 nodes). on each node of the cluster gtid_mode is on with all the other settings for enabling GTID. I have another stand alone instance (GTID enabled) that is replicating in a master/master setup from one node of XtraDB cluster. Basically data flows from XtraDB cluster node to stand alone node and the oterh way around. It works fine until the XtraDB node is going down. At this stage I would like to point the stand alone instance to replicate from another node. If I just run the change master to master_host=‘second_node’, MASTER_AUTO_POSITION=1 it may fail as it doesn’t find all the binary logs. I can always run a reset master on the stand alone instance (replicated from the XtraDB cluster) and set GTID_PURGED to whatever gtid_executed shows on the new master node.
But this would break replication from the node to xtradb cluster as the cluster node will not find the binary logs on the stand alone node.
Has anyone tried this setup?