irr123 / clickhousecluster_dockercompose Goto Github PK

There are three shard by one replicas to discover cluster with Distributed-tables.

Install Docker Engine and Docker Compose.

Call docker-compose up -d --remove-orphans.

In result you will have:

• haproxy
• Bulk
• 3 Zookeeper
• 3 Clickhouse

Next connect to any clickhouse instance by clickhouse-client --host=127.0.0.1 --port=9000 --multiline and create regular table at first:

CREATE TABLE IF NOT EXISTS default.test_real ON CLUSTER my_cluster (
  date DateTime DEFAULT now(),
  uuid String,
  data String,
)
ENGINE = MergeTree()
TTL date + INTERVAL 90 DAY
PARTITION BY toYYYYMMDD(date)
ORDER BY (uuid);

And distributed next:

CREATE TABLE IF NOT EXISTS default.test ON CLUSTER my_cluster AS default.test_real
ENGINE = Distributed(my_cluster, default, test_real, halfMD5(uuid));

So now you have two main ports:

• http port 8124 for inserting rows into test_real-table
• tcp port 9000 for selecting rows from test-table

You may insert by single rows, bulk will automatically aggregate it and write bulk into instance with round-robin sharding.

Select query will be distribute on all cluster.