Spark App(用Spark APIs编写的)需要submit到Spark Cluster运行,对于Scala编写的代码,提交之前要用sbt或者maven把以下内容:
全部打包成一个大的jar文件,这样代码就不会因为没有依赖无法在集群中运行。
我司用Scala编写Spark streaming应用,实现读取Kafka数据,处理后存储到cassandra集群中。这里需要用到一个包spark-streaming-kafka
,之前用的spark1.6.0
的版本。sbt中的配置如下:
libraryDependencies ++= Seq(
// Spark dependency
"com.eaio.uuid" % "uuid" % "3.2",
"org.apache.spark" %% "spark-core" % "1.6.0" % "provided",
"org.apache.spark" %% "spark-sql" % "1.6.0" % "provided",
"org.apache.spark" %% "spark-streaming" % "1.6.0" % "provided",
"org.apache.spark" %% "spark-streaming-kafka" % "1.6.0",
"com.datastax.spark" %% "spark-cassandra-connector" % "1.6.0-M2",
// Third-party libraries
"com.github.scopt" %% "scopt" % "3.4.0"
)
升级到Spark 2.0.0后需要更新软件包版本,于是将sbt构建配置中的依赖部分改为:
libraryDependencies ++= Seq(
// Spark dependency
"com.eaio.uuid" % "uuid" % "3.2",
"org.apache.spark" %% "spark-core" % "2.0.0" % "provided",
"org.apache.spark" %% "spark-sql" % "2.0.0" % "provided",
"org.apache.spark" %% "spark-streaming" % "2.0.0" % "provided",
"org.apache.spark" %% "spark-streaming-kafka" % "2.0.0",
"com.datastax.spark" %% "spark-cassandra-connector" % "2.0.0-M2",
// Third-party libraries
"com.github.scopt" %% "scopt" % "3.4.0"
)
本以为这样修改后重新构建就没问题了。但是我太天真了,构建后报错,提示:
[warn] ::::::::::::::::::::::::::::::::::::::::::::::
[warn] :: UNRESOLVED DEPENDENCIES ::
[warn] ::::::::::::::::::::::::::::::::::::::::::::::
[warn] :: org.apache.spark#spark-streaming-kafka_2.10;2.0.0: not found
[warn] ::::::::::::::::::::::::::::::::::::::::::::::
[warn]
[warn] Note: Unresolved dependencies path:
[warn] org.apache.spark:spark-streaming-kafka_2.10:2.0.0 (/home/linker/workspace/linkerwp/linkerStreaming/build.sbt#L12-23)
[warn] +- Linker Networks Inc.:linker-streaming_2.10:0.0.1
sbt.ResolveException: unresolved dependency: org.apache.spark#spark-streaming-kafka_2.10;2.0.0: not found
说明这个包是不存在的,于是马上到maven repo上去找。常见的Maven公共repo有:
进入网址输入spark-streaming-kafka
搜索后数来好几和选项,前面4个结果都是不支持Spark 2.0.0的,这让我误以为Spark 2.0.0还不支持Kafka,这个想法被前面几个搜索结果误导了。因为对于2.0.0的Spark,Kafka添加了具体的版本号!spark-streaming-kafka
→spark-streaming-kafka-0-8
就可以找到了(实际上这个版本也在maven repo的搜索结果,因为靠后我没有去看)!!