ʹÓÃMaxwell+Python+ClickHouseʵÏÖÊý¾Ýչʾ

# ÈÕÖ¾¼¶±ð
log_level=info
client_id=12345
replica_server_id=12345

# Maxwell ÔªÊý¾ÝÊý¾Ý¿âÐÅÏ¢
host=192.168.100.1
port=3306
user=root
password=123456

#producer kafkaµÄÅäÖÃÐÅÏ¢
producer=kafka
kafka.bootstrap.servers=localhost:9092
kafka_topic=binlog_service_A
kafka_partition_hash=murmur3
producer_partition_by=primary_key

#±»Í¬²½µÄÊý¾Ý¿âµÄÐÅÏ¢
replication_host=192.168.100.11
replication_user=dts
replication_password=dts
replication_port=3306
jdbc_options=useSSL=false&serverTimezone=Asia/Shanghai

# ÅäÖùýÂËÆ÷£¬¿ÉÒÔÅäÖÃֻͬ²½Ö¸¶¨µÄÊý¾Ý¿â£¬Ö¸¶¨µÄ±íµÄÊý¾Ý±ä¶¯¡£
#filter=exclude: *.*

# -*- coding: utf-8 -*-

# MySQL Settings
MYSQL_SETTINGS = {
    "host": "192.168.100.11",
    "port": 3306,
    "user": "dts",
    "passwd": "dts"
}

#
MySQL_CLUSTER_NAME = "·þÎñ-A-RDS"

# Kafka Settings
bootstrap_servers = '127.0.0.1:9092'
original_topic = "binlog_service_A"
washed_topic = "binlog_washed"

# -*- coding: utf-8 -*-
import json
from kafka import KafkaConsumer, KafkaProducer
import configs

consumer_config = {
    'bootstrap_servers': configs.bootstrap_servers,
    'group_id': 'my_consumer_group',
    'auto_offset_reset': 'latest',  # ¿ÉÑ¡ earliest latest  ĬÈÏÆô¶¯ºó´Ó×îеÄλÖÿªÊ¼Ïû·Ñ
    'enable_auto_commit': True
}
producer_config = {
    'bootstrap_servers': configs.bootstrap_servers
}


# ´´½¨KafkaÏû·ÑÕߺÍÉú²úÕß
consumer = KafkaConsumer(configs.original_topic, **consumer_config)
producer = KafkaProducer(**producer_config)

# Ñ­»·Ïû·ÑÏûÏ¢²¢´¦Àí
for message in consumer:
    data = message.value.decode('utf-8')
    # print(data)

    database = json.loads(data).get('database')
    table = json.loads(data).get('table')
    type = json.loads(data).get('type')
    ts = json.loads(data).get('ts')
    xid = json.loads(data).get('xid')
    xoffset = json.loads(data).get('xoffset')

    result = {}
    result['database'] = database
    result['table'] = table
    result['type'] = type
    result['ts'] = ts
    result['xid'] = xid
    result['xoffset'] = xoffset
    result['cluster'] = configs.MySQL_CLUSTER_NAME

    res = json.dumps(result, ensure_ascii=False)

    new_data = res
    print(new_data)

    # ·¢ËÍ´¦ÀíºóµÄÏûÏ¢
    producer.send(configs.washed_topic, value=new_data.encode('utf-8'))

# ¹Ø±ÕKafkaÏû·ÑÕߺÍÉú²úÕßÁ¬½Ó
consumer.close()
producer.close()

{
    clickhouse: {
        hosts: [
            [
                127.0.0.1
            ]
        ]
        port: 9000
        db: default
        username: ""
        password: ""
        retryTimes: 0
    }
    kafka: {
        brokers: 127.0.0.1:9092
    }
    task: {
        name: binlog_washed
        topic: binlog_washed
        consumerGroup: clickhouse_sinker
        earliest: false
        parser: json
        autoSchema: true
        tableName: binlog_washed
        excludeColumns: []
        bufferSize: 100000
        "dynamicSchema": {
            "enable": true,
            "maxDims": 1024,
            "whiteList": "^[0-9A-Za-z_]+$",
            "blackList": "@"
        },
    }
    logLevel: info
}

# -*- encoding: utf-8 -*-
# Ïà¹ØµÄ¶¨Ê±ÈÎÎñ£¬¶¼·Åµ½ÕâÀï

from datetime import date, timedelta
from clickhouse_driver import Client

from apscheduler.schedulers.background import BlockingScheduler
from apscheduler.triggers.cron import CronTrigger
from apscheduler.executors.pool import ThreadPoolExecutor


def ck_job():
    client = Client('localhost')

    calc_date = (date.today() - timedelta(days=1)).strftime('%Y-%m-%d')
    purge_date = (date.today() - timedelta(days=30)).strftime('%Y-%m-%d')

    # 1 ¸ù¾Ýdayͳ¼Æ
    insert_sql = """INSERT INTO binlog_washed_by_day SELECT toDate(ts) AS day,type,count(*) AS count,cluster FROM binlog_washed where toDate(ts) = """ + "'" + calc_date + "'" + """
    GROUP BY day,type,cluster ORDER BY day ASC,type ASC;"""

    res = client.execute(insert_sql)
    print(res)

    # 2 ¸ù¾Ýtable dayͳ¼Æ
    insert_sql = """INSERT INTO binlog_washed_by_table SELECT toDate(ts) AS day, database, table, type, count(*) AS count, cluster FROM binlog_washed where toDate(ts) = """ + "'" + calc_date + "'" + """GROUP BY day, database,table,type,cluster ORDER BY day ASC, type ASC"""

    res = client.execute(insert_sql)
    print(res)

    # 3 ɾÔçÆÚ·ÖÇø
    alter_sql = """ALTER TABLE binlog_washed DROP PARTITION """ + "'" + purge_date + "'"

    res = client.execute(alter_sql)
    print(res)

    client.disconnect()


if __name__ == "__main__":

    executors = {"default": ThreadPoolExecutor(1)}

    scheduler = BlockingScheduler(executors=executors)

    # ÿÌìµÄ03:50:01Ö´ÐÐ×÷Òµ
    intervalTrigger=CronTrigger(hour=3, minute=50, second=1)

    scheduler.add_job(ck_job, intervalTrigger, id="ck_job_id")
    scheduler.start()

-- Ã÷ϸ±í
CREATE TABLE default.binlog_washed
(
    `ts` DateTime,
    `database` String,
    `table` String,
    `type` String,
    `xid` UInt8,
    `xoffset` UInt8,
    `cluster` String
)
ENGINE = MergeTree
PARTITION BY toDate(ts)
ORDER BY (ts, database, table)
TTL ts + toIntervalMonth(30)
SETTINGS index_granularity = 8192;


-- ͳ¼Æ±í
CREATE TABLE default.binlog_washed_by_day
(
    `day` Date,
    `type` String,
    `count` UInt64,
    `cluster` String
)
ENGINE = MergeTree
PARTITION BY day
ORDER BY (day, type, cluster)
TTL day + toIntervalMonth(30)
SETTINGS index_granularity = 8192;


CREATE TABLE default.binlog_washed_by_table
(
    `day` Date,
    `database` String,
    `table` String,
    `type` String,
    `count` UInt64,
    `cluster` String
)
ENGINE = MergeTree
PARTITION BY day
ORDER BY (day, database, table, type, cluster)
TTL day + toIntervalMonth(30)
SETTINGS index_granularity = 8192;

# supervisorctl status 
ck_jobs                          RUNNING   pid 31499, uptime 19 days, 18:13:12  -- ¹²ÓÃ
ck_sinker                        RUNNING   pid 2813, uptime 15 days, 23:03:46 -- ¹²ÓÃ
cluster1_maxwell                 RUNNING   pid 8024, uptime 19 days, 2:52:00  -- ÿÌ×¼¯Èº¶¼ÒªÆô¶¯
cluster1_wash_binlog             RUNNING   pid 5252, uptime 19 days, 2:54:24  -- ÿÌ×¼¯Èº¶¼ÒªÆô¶¯
cluster2_maxwell                 RUNNING   pid 31896, uptime 19 days, 2:55:21
cluster2_wash_binlog             RUNNING   pid 4495, uptime 19 days, 2:56:43