When a DDL is issued in RSU it can timeout due to ongoing commits
In the log we see
Which comes from sql/wsrep_mysqld.cc (line 1925 for 5.7.18)
Is there any reason this can not be retried before giving up?
Introducing a variable wsrep_rsu_ddl_retry and wsrep_rsu_ddl_retry_sleep would allow users to tune for how long they want to wait before giving up. Alternatively make the 5000ms configurable so the timeout would be longer and users can adjust according to their needs and workloads.