Reported on Codership mailinglist.

Hi,I'm having problems with MariaDB Galera Cluster 10.0.16.

MariaDB version: 10.0.16 (from RHEL 6 repo)
OS: RHEL 6.5

Every time "create table T2 as select * from T1" is executed on one node all the other nodes will fail with:
[ERROR] Slave SQL: Error executing row event: 'Table 'dbname.T2' doesn't exist', Internal MariaDB error code: 1146

And all nodes except the one which the statement was executed on is shutdown:
[ERROR] WSREP: Node consistency compromized, aborting...

Leaving me with just 1 node left running.
There are some similar bugs related to "create as select", but they should have been fixed in 10.0.15/5.5.41.

The problem is simple to reproduce:
1. Create a cluster with 2 or more nodes.

2. Execute the following in any database on one of the nodes:
create table test(id int Primary Key, name varchar(10));
insert into table test values(1, "Test");
create table test2 as select * from test;

Does any one else see this or could it be due to a config error on my side?

Error log from the other two nodes:

150212 8:52:11 [Warning] WSREP: BF applier failed to open_and_lock_tables: 1146, fatal: 0 wsrep = (exec_mode: 1 conflict_state: 0 seqno: 3)
150212 8:52:11 [ERROR] Slave SQL: Error executing row event: 'Table 'test.test2' doesn't exist', Internal MariaDB error code: 1146
150212 8:52:11 [Warning] WSREP: RBR event 2 Write_rows_v1 apply warning: 1146, 3
150212 8:52:11 [Warning] WSREP: Failed to apply app buffer: seqno: 3, status: 1
at galera/src/trx_handle.cpp:apply():351