大致思路：
canal去mysql拉取数据，放在canal所在的节点上，并且自身对外提供一个tcp服务，我们只要写一个连接该服务的客户端，去拉取数据并且指定往kafka写数据的格式就能达到以protobuf的格式往kafka中写数据的要求。
1. 配置canal（/bigdata/canal/conf/canal.properties），然后启动canal，这样就会开启一个tcp服务
2. 写拉取数据的客户端代码
PbOfCanalToKafka
package cn._51doit.flink.canal;
import cn._51doit.proto.OrderDetailProto;
import com.alibaba.google.common.base.CaseFormat;
import com.alibaba.otter.canal.client.CanalConnector;
import com.alibaba.otter.canal.client.CanalConnectors;
import com.alibaba.otter.canal.protocol.CanalEntry;
import com.alibaba.otter.canal.protocol.Message;
import org.apache.kafka.clients.producer.KafkaProducer;
import org.apache.kafka.clients.producer.ProducerRecord; import java.net.InetSocketAddress;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Properties; public class PbOfCanalToKafka {
public static void main(String[] args) throws Exception {
CanalConnector canalConnector = CanalConnectors.newSingleConnector((new InetSocketAddress("192.168.57.12", 11111)), "example", "canal", "canal123");
// 1 配置参数
Properties props = new Properties();
//连接kafka节点
props.setProperty("bootstrap.servers", "feng05:9092,feng06:9092,feng07:9092");
props.setProperty("key.serializer", "org.apache.kafka.common.serialization.StringSerializer");
props.setProperty("value.serializer", "org.apache.kafka.common.serialization.ByteArraySerializer");
KafkaProducer<String, byte[]> producer = new KafkaProducer<String, byte[]>(props); while (true) {
//建立连接
canalConnector.connect();
//订阅bigdata数据库下的所有表
canalConnector.subscribe("doit.orderdetail");
//每100毫秒拉取一次数据
Message message = canalConnector.get(10);
if (message.getEntries().size() > 0) {
// System.out.println(message);
List<CanalEntry.Entry> entries = message.getEntries();
for (CanalEntry.Entry entry : entries) {
//获取表名
String tableName = entry.getHeader().getTableName();
CanalEntry.RowChange rowChange = CanalEntry.RowChange.parseFrom(entry.getStoreValue());
List<CanalEntry.RowData> rowDatasList = rowChange.getRowDatasList();
// System.out.println(rowDatasList);
//判断对数据库操作的类型，这里只采集INSERT/update的数据
OrderDetailProto.OrderDetail.Builder bean = OrderDetailProto.OrderDetail.newBuilder();
CanalEntry.EventType eventType = rowChange.getEventType();
if (eventType == CanalEntry.EventType.INSERT || eventType == CanalEntry.EventType.UPDATE) {
for (CanalEntry.RowData rowData : rowDatasList) {
List<CanalEntry.Column> afterColumnsList = rowData.getAfterColumnsList();
System.out.println("======================打印afterColumnsList==============================");
System.out.println(afterColumnsList);
Map<String, String> kv = new HashMap<String, String>();
for (CanalEntry.Column column : afterColumnsList) {
String propertyName = CaseFormat.LOWER_UNDERSCORE.to(CaseFormat.LOWER_CAMEL, column.getName());
kv.put(propertyName, column.getValue());
}
// 设置属性
bean.setAmount(Integer.parseInt(kv.get("amount")));
bean.setMoney(Double.parseDouble(kv.get("money")));
bean.setOrderId(Long.parseLong(kv.get("orderId")));
bean.setCreateTime(kv.get("createTime"));
bean.setUpdateTime(kv.get("updateTime"));
bean.setId(Integer.parseInt(kv.get("id")));
bean.setSku(Long.parseLong(kv.get("sku")));
bean.setCategoryId(Integer.parseInt(kv.get("categoryId")));
//将数据转成JSON格式，然后用Kafka的Producer发送出去
byte[] bytes = bean.build().toByteArray();
ProducerRecord<String, byte[]> record = new ProducerRecord<>(tableName, bytes);
producer.send(record);
}
}
}
}
}
}
}
注意：数据被拉取到canal的格式不为json（若是不开启tcp服务，直接将数据发送给kafka，则数据在kafka中的格式为json），OrderDetailProto的生成见flink实时项目day07
Message
Message[id=1,entries=[header {
version: 1
logfileName: "mysql-bin.000002"
logfileOffset: 6669
serverId: 1
serverenCode: "UTF-8"
executeTime: 1594134782000
sourceType: MYSQL
schemaName: ""
tableName: ""
eventLength: 31
}
entryType: TRANSACTIONEND
storeValue: "\022\0042179"
, header {
version: 1
logfileName: "mysql-bin.000002"
logfileOffset: 6765
serverId: 1
serverenCode: "UTF-8"
executeTime: 1594147469000
sourceType: MYSQL
schemaName: ""
tableName: ""
eventLength: 80
}
entryType: TRANSACTIONBEGIN
storeValue: " A"
, header {
version: 1
logfileName: "mysql-bin.000002"
logfileOffset: 6911
serverId: 1
serverenCode: "UTF-8"
executeTime: 1594147469000
sourceType: MYSQL
schemaName: "doit"
tableName: "orderdetail"
eventLength: 82
eventType: INSERT
props {
key: "rowsCount"
value: "1"
}
}
entryType: ROWDATA
storeValue: "\b\177\020\001P\000b\332\002\022\'\b\000\020\373\377\377\377\377\377\377\377\377\001\032\002id \001(\0010\000B\00212R\nbigint(20)\0220\b\001\020\373\377\377\377\377\377\377\377\377\001\032\border_id \000(\0010\000B\00529002R\nbigint(20)\022#\b\002\020\004\032\vcategory_id \000(\0010\000B\0012R\aint(11)\022#\b\003\020\f\032\003sku \000(\0010\000B\00520001R\vvarchar(50)\022!\b\004\020\b\032\005money \000(\0010\000B\0062000.0R\006double\022\036\b\005\020\004\032\006amount \000(\0010\000B\0012R\aint(11)\0227\b\006\020]\032\vcreate_time \000(\0010\000B\0232020-07-01 20:19:08R\ttimestamp\0227\b\a\020]\032\vupdate_time \000(\0010\000B\0232020-07-02 20:19:13R\ttimestamp"
, header {
version: 1
logfileName: "mysql-bin.000002"
logfileOffset: 6993
serverId: 1
serverenCode: "UTF-8"
executeTime: 1594147469000
sourceType: MYSQL
schemaName: ""
tableName: ""
eventLength: 31
}
entryType: TRANSACTIONEND
storeValue: "\022\0042197"
],raw=false,rawEntries=[]]
在mysql表orderdetail表中添加了一行
rowDatasList
[afterColumns {
index: 0
sqlType: -5
name: "id"
isKey: true
updated: true
isNull: false
value: "13"
mysqlType: "bigint(20)"
}
afterColumns {
index: 1
sqlType: -5
name: "order_id"
isKey: false
updated: true
isNull: false
value: "29002"
mysqlType: "bigint(20)"
}
afterColumns {
index: 2
sqlType: 4
name: "category_id"
isKey: false
updated: true
isNull: false
value: "3"
mysqlType: "int(11)"
}
afterColumns {
index: 3
sqlType: 12
name: "sku"
isKey: false
updated: true
isNull: false
value: "22333"
mysqlType: "varchar(50)"
}
afterColumns {
index: 4
sqlType: 8
name: "money"
isKey: false
updated: true
isNull: false
value: "1111.0"
mysqlType: "double"
}
afterColumns {
index: 5
sqlType: 4
name: "amount"
isKey: false
updated: true
isNull: false
value: "3"
mysqlType: "int(11)"
}
afterColumns {
index: 6
sqlType: 93
name: "create_time"
isKey: false
updated: true
isNull: false
value: "2020-07-01 22:02:50"
mysqlType: "timestamp"
}
afterColumns {
index: 7
sqlType: 93
name: "update_time"
isKey: false
updated: true
isNull: false
value: "2020-07-02 22:02:54"
mysqlType: "timestamp"
}
]
afterColumnsList
[index: 0
sqlType: -5
name: "id"
isKey: true
updated: false
isNull: false
value: "12"
mysqlType: "bigint(20)"
, index: 1
sqlType: -5
name: "order_id"
isKey: false
updated: false
isNull: false
value: "29002"
mysqlType: "bigint(20)"
, index: 2
sqlType: 4
name: "category_id"
isKey: false
updated: false
isNull: false
value: "2"
mysqlType: "int(11)"
, index: 3
sqlType: 12
name: "sku"
isKey: false
updated: true
isNull: false
value: "20011"
mysqlType: "varchar(50)"
, index: 4
sqlType: 8
name: "money"
isKey: false
updated: false
isNull: false
value: "2000.0"
mysqlType: "double"
, index: 5
sqlType: 4
name: "amount"
isKey: false
updated: false
isNull: false
value: "2"
mysqlType: "int(11)"
, index: 6
sqlType: 93
name: "create_time"
isKey: false
updated: false
isNull: false
value: "2020-07-01 20:19:08"
mysqlType: "timestamp"
, index: 7
sqlType: 93
name: "update_time"
isKey: false
updated: false
isNull: false
value: "2020-07-02 20:19:13"
mysqlType: "timestamp"
]
3. 若是想从kafka中读取protobuf格式的数据，则需要自定义序列化器，这里以flink读取盖格师的数据为例
具体见flink实时项目day07
