how to change Mysql 5.5 slave to PXC 5.6 node

Hello,

I have mysql master-slave cluster (multiple slaves), and I want to migrate seamlessly to 5.6 xtraDB cluster, so i decided that proper way would be turn one of the slaves to first node of PXC cluster, while still maintaining replication, this was mentioned in webinar ([url]https://www.percona.com/files/presentations/WEBINAR-Migrating-to-Xtradb-Cluster.pdf[/url]). What is my problem? I couldn find solution how to turn directly (or indirectly) 5.5 mysql to 5.6 XtraDB and thet based on this node create rest of the cluster.
Is proper way to upgrade this slave to 5.6 mysql an then migrate to XtraDB, or there is direct method?
Maybe you can help me?

Regards,

So you have a master-slave replication cluster now, and you want to migrate to PXC 5.6. First upgrade the slave to latest 5.5 version of Percona Server (optional but would be good to perform this step). On one of the slave nodes do the following:


slave# service mysql stop
slave# rpm -qa| grep Percona-Server| xars rpm -e --nodeps
slave# yum install Percona-XtraDB-Cluster-server-56 Percona-XtraDB-Cluster-client-56
slave# service mysql start
slave# mysql -e 'SHOW SLAVE STATUS\G'

At this point you should be running PXC without the wsrep replication related configuration done.

Configure the node (old slave) with appropriate wsrep settings then bootstrap the node. MySQL should resume replication on the newly bootstrapped node.

Do the same steps above to migrate other slaves to PXC but this time instead of reconnecting the slave to the master make sure to stop slave replication, reset slave, configure wsrep settings and then start mysql so it will connect to the bootstrapped node and not replicate from the master. Once node2 is up and in “SYNC” with node1 you can point your applications to node1. Stop replication on node1 to master, then reset slave replication.

On old master, remove old binaries and install PXC. Configure wsrep settings to be identical to nodes 1 and 2. Start it up and SST/IST from node2 (–wsrep_donor_node) and you should have a 3node PXC cluster after this barring other problems along the way.

Thank you, i have other issues, but i’ll create new topic.
Regards :slight_smile: