Standup a CS cluster as usual (i used the standard https://github.com/mariadb-corporation/columnstore-ansible-aws 3 node, s3 deployment)
# watch GTID replication of all nodes with maxscale in window 1
watch -n1 "maxctrl list servers"

# watch the number of CS tables be created on PM1 in window 2
watch -n1 " mariadb -e \"select table_schema, engine, count(*) as quantity from information_schema.tables where table_schema not in ('information_schema','mysql','performance_schema') group by table_schema, engine ;\" "

# watch the number of CS tables be created on PM2 in window 3
watch -n1 " mariadb -e \"select table_schema, engine, count(*) as quantity from information_schema.tables where table_schema not in ('information_schema','mysql','performance_schema') group by table_schema, engine ;\" "

# Run the script of DDL creation or run any create tables with session variable sql_log_bin=0
mariadb < massSchemas.sql
# Notice GTID doesnt move, slaves cant query this table - they only exist on the master

# Notice masstest.affiliato doesnt exist on slaves 
# Notice cpimport doesnt incerement or break GTID replication
cpimport masstest affiliato /tmp/massCp.csv -s ','

# ERROR TIME #
# But using LDIF increments the GTID breaking the slaves from ever catching up without manual intervention
mariadb masstest -e "LOAD DATA LOCAL INFILE '/tmp/massCp.csv' INTO TABLE affiliato FIELDS TERMINATED BY ',';"

# specifc DDL not important but sharing
CREATE TABLE `affiliato` (
  `af_id` int(11) DEFAULT NULL,
  `af_desc` varchar(40) DEFAULT NULL,
  `af_in_id` int(11) DEFAULT NULL
) ENGINE=Columnstore DEFAULT CHARSET=utf8;