Cluster not starting when mysqld_multi is used

I have a Percona cluster.
For a new application I needed to run a non-clustered instance of mysqld on one node.
So I created a separate configuration in my.cnf, and everything worked great.

However, I decided to check if it runs after reboot, and found out that it does not.
Mysqld starts, but it doesn’t try to access the cluster, despite all the setting regarding the clustered instance remained the same.

Here is the startup log:

150202 10:47:48 mysqld_safe Starting mysqld daemon with databases from /var/lib/mysql
150202 10:47:48 mysqld_safe Skipping wsrep-recover for 646956ec-a7c6-11e4-a49a-0eafe484ab9f:35 pair
150202 10:47:48 mysqld_safe Assigning 646956ec-a7c6-11e4-a49a-0eafe484ab9f:35 to wsrep_start_position
150202 10:47:48 [Note] WSREP: wsrep_start_position var submitted: ‘646956ec-a7c6-11e4-a49a-0eafe484ab9f:35’
150202 10:47:48 [Note] Plugin ‘FEDERATED’ is disabled.
150202 10:47:48 InnoDB: The InnoDB memory heap is disabled
150202 10:47:48 InnoDB: Mutexes and rw_locks use GCC atomic builtins
150202 10:47:48 InnoDB: Compressed tables use zlib 1.2.3
150202 10:47:48 InnoDB: Using Linux native AIO
150202 10:47:48 InnoDB: Initializing buffer pool, size = 128.0M
150202 10:47:49 InnoDB: Completed initialization of buffer pool
150202 10:47:49 InnoDB: highest supported file format is Barracuda.
150202 10:47:49 InnoDB: Waiting for the background threads to start
150202 10:47:50 Percona XtraDB ( 5.5.39-36.0 started; log sequence number 1600827
150202 10:47:50 [Note] Event Scheduler: Loaded 0 events
150202 10:47:50 [Note] WSREP: Initial position: 646956ec-a7c6-11e4-a49a-0eafe484ab9f:35
150202 10:47:50 [Note] WSREP: wsrep_load(): loading provider library ‘none’
150202 10:47:50 [Note] /usr/sbin/mysqld: ready for connections.
Version: ‘5.5.39-36.0-55’ socket: ‘/var/lib/mysql/mysql.sock’ port: 3306 Percona XtraDB Cluster (GPL), Release rel36.0, Revision 824, WSREP version 25.11, wsrep_25.11.r4023

If I return back to the single-instance my.cnf then this node joins cluster again.
Does myslqd_multi support clustered instances at all? And if it does, which option should I add?

I’ve switched to mysql sandboxes and it’s ok, but still interesting to know why mysqld_mutli failed for me

You may like reading this: