You bootsrap the first node, next you start the other servers at that point they will all be members of the same cluster. At that point you stop the ‘bootstarp’ node with systemctl stop mysql@bootstrap then start it again but using the normal systemctl start for percona and the cluster should be up and running normally.
I experience the same issue on Debian 9. After starting node 1 as bootstrap (sudo /etc/init.d/mysql bootstrap-pxcand joining node 2 and node 3, node 1 will not stop with /etc/init.d/mysql stop which apparently does nothing. I did a standard kill on the mysql processes and it shutdown cleanly. At that point I was able to do a standard node 1 start and all is well. But it is odd that mysql stop does not work after bootstrap start.
It’s strange, I’m install PXC “apt-get install percona-xtradb-cluster-server” through its official repo: “wget https://repo.percona.com/apt/percona-release_latest.$(lsb_release -sc)_all.deb && dpkg -i … && percona-release setup pxc57”.
And the installation progress successfully without error, the mysqld version is: “mysqld Ver 5.7.27-30-57 for debian-linux-gnu on x86_64 (Percona XtraDB Cluster (GPL), Release rel30, Revision 64987d4, WSREP version 31.39, wsrep_31.39)”.
All looks fine, BUT there is no systemd configurations (neither “/lib/systemd/systemd/mysql.service” nor “/lib/systemd/systemd/mysql@.service”). Only a sysv “/etc/init.d/mysql” script has created.