需要注意的是,flink連接es時,端口號需要使用tcp端口(一般es默認http端口爲9200,tcp端口爲9300)。
代碼如下:
//es配置
val config = new java.util.HashMap[String, String]
config.put("bulk.flush.max.actions", "1")
config.put("cluster.name", "your_name")
val transportAddresses = new java.util.ArrayList[InetSocketAddress]
transportAddresses.add(new InetSocketAddress(InetAddress.getByName("host1"), 9300))
transportAddresses.add(new InetSocketAddress(InetAddress.getByName("host2"), 9300))
transportAddresses.add(new InetSocketAddress(InetAddress.getByName("host3"), 9300))
transportAddresses.add(new InetSocketAddress(InetAddress.getByName("host4"), 9300))
transportAddresses.add(new InetSocketAddress(InetAddress.getByName("host5"), 9300))
resultStream.addSink(new ElasticsearchSink(config, transportAddresses, new ElasticsearchSinkFunction[(String, Int, Int, Long, Long)] {
def createIndexRequest(element: (String, Int, Int, Long, Long)): IndexRequest = {
val dataMap = new java.util.HashMap[String, Any]
val keys = element._1.split("_")
val date = keys(0)
val productId = keys(1)
val channelId = keys(2)
val carriorType = keys(3)
val provinceName = keys(4)
dataMap.put("date", date)
dataMap.put("productId", productId)
dataMap.put("channelId", channelId)
dataMap.put("carriorType", carriorType)
dataMap.put("provinceName", provinceName)
dataMap.put("pv", element._2)
dataMap.put("uv", element._3)
val idCompact: String = date + productId + channelId + carriorType + provinceName
val id: String = DataUtils.simpleEncode(idCompact)
return Requests.indexRequest()
.index("your_index")
.`type`("your_type")
.id(id)
.source(dataMap)
}
def process(element: (String, Int, Int, Long, Long), ctx: RuntimeContext, indexer: RequestIndexer): Unit = {
indexer.add(createIndexRequest(element))
}
}))
pom文件:
<dependency> <groupId>org.apache.flink</groupId> <artifactId>flink-connector-elasticsearch5_2.10</artifactId> <version>1.3.3</version> </dependency>