spark 怎么读写 elasticsearch

参考这两篇文章：

https://www.bmc.com/blogs/spark-elasticsearch-hadoop/
https://blog.pythian.com/updating-elasticsearch-indexes-spark/

 

Pre-requisite:

先装上 elasticsearch-hadoop 包

 

Step-by-Step guide

 

1. 先在ES创建3个document

[mshuai@node1 ~]$ curl -XPUT --header 'Content-Type: application/json' http://your_ip:your_port/school/doc/1 -d '{
   "school" : "Clemson"
}'
[mshuai@node1 ~]$ curl -XPUT --header 'Content-Type: application/json' http://your_ip:your_port/school/doc/2 -d '{
   "school" : "Harvard"
}'

 

2. Spark 里面去读，这里是pyspark 代码

 

reader = spark.read.format("org.elasticsearch.spark.sql").option("es.read.metadata", "true").option("es.nodes.wan.only","true").option("es.port","your_port").option("es.net.ssl","false").option("es.nodes", "your_ip")
df = reader.load("school")
df.show()

输出这个格式的信息

 

3. 接下来尝试update 一个记录，先得到一个要改的id

 

hot = df.filter(df["school"] == 'Harvard') \
        .select(expr("_metadata._id as id")).withColumn('hot', lit(True))
hot.show()

 

 

4. 先来加一列

 

esconf={}
esconf["es.mapping.id"] = "id"
esconf["es.mapping.exclude"]='id'
esconf["es.nodes"] = "your_ip"
esconf["es.port"] = "your_port"
esconf["es.write.operation"] = "update"
esconf["es.nodes.discovery"] = "false"
esconf["es.nodes.wan.only"] = "true"

hot.write.format("org.elasticsearch.spark.sql").options(**esconf).mode("append").save("school/doc")

看，成功加了 hot 列，满足条件的记录赋了对应的值

  

 

5. 又来update已经存在的信息

 

esconf={}
esconf["es.mapping.id"] = "id"
esconf["es.nodes"] = "your_ip"
esconf["es.port"] = "your_port"
esconf["es.update.script.inline"] = "ctx._source.school = params.school"
esconf["es.update.script.params"] = "school:"
esconf["es.write.operation"] = "update"
esconf["es.nodes.discovery"] = "false"
esconf["es.nodes.wan.only"] = "true"
hot.write.format("org.elasticsearch.spark.sql").options(**esconf).mode("append").save("school/doc")
reader.load("school").show()

嗯。。。成功把Harvard改成了川大

 

另外，怎么upload attachment 到ES呢？可以用这个plugin Ingest Attachment Processor Plugin

 

END