apache-spark - 如何使用 AWS ECS Fargate 而不是 EMR/EC2 运行 apache spark 作业？

Question

AWS Elastic MapReduce 有很多功能，但它有一些粗糙的边缘，我想回避一些我想在 Apache Spark 中进行的相当便宜的计算。具体来说，我想看看是否可以在 AWS ECS/Fargate 上运行（scala）火花应用程序。如果我能让它只与一个在客户端/本地模式下运行的容器一起工作，那就更好了。

我首先使用 hadoop3（用于 AWS STS 支持）和选择的 kubernetes 配置文件分发 Spark：

# in apache/spark git repository under tag v2.4.0
./dev/make-distribution.sh --name hadoop3-kubernetes -Phadoop-3.1 -Pkubernetes -T4

然后从该发行版中构建一个通用的 spark docker 映像：

docker build -t spark:2.4.0-hadoop3.1 -f kubernetes/dockerfiles/spark/Dockerfile .

然后在我的项目中，我在 tope 上构建了另一个 docker 映像，将我的 sbt-assembled uberjar 复制到工作目录中，并将入口点设置为 spark-submit shell 脚本。

# Dockerfile
FROM spark:2.4.0-hadoop3.1
COPY target/scala-2.11/my-spark-assembly.jar .
ENTRYPOINT [ "/opt/spark/bin/spark-submit" ]

在我的本地机器上，我可以通过在 docker-compose 命令规范中提供适当的参数来运行该应用程序：

# docker-compose.yml
...
   command:
     - --master
     - local[*]
     - --deploy-mode
     - client
     - my-spark-assembly.jar

不幸的是，在 Fargate ECS 中，我通过以下堆栈跟踪写入 CloudWatch 快速失败：

Exception in thread "main" java.lang.ExceptionInInitializerError
at org.apache.spark.SparkConf$.<init>(SparkConf.scala:714)
at org.apache.spark.SparkConf$.<clinit>(SparkConf.scala)
at org.apache.spark.SparkConf$$anonfun$getOption$1.apply(SparkConf.scala:388)
at org.apache.spark.SparkConf$$anonfun$getOption$1.apply(SparkConf.scala:388)
at scala.Option.orElse(Option.scala:289)
at org.apache.spark.SparkConf.getOption(SparkConf.scala:388)
at org.apache.spark.SparkConf.get(SparkConf.scala:250)
at org.apache.spark.deploy.SparkHadoopUtil$.org$apache$spark$deploy$SparkHadoopUtil$$appendS3AndSparkHadoopConfigurations(SparkHadoopUtil.scala:463)
at org.apache.spark.deploy.SparkHadoopUtil$.newConfiguration(SparkHadoopUtil.scala:436)
at org.apache.spark.deploy.SparkSubmit$$anonfun$2.apply(SparkSubmit.scala:334)
at org.apache.spark.deploy.SparkSubmit$$anonfun$2.apply(SparkSubmit.scala:334)
at scala.Option.getOrElse(Option.scala:121)
at org.apache.spark.deploy.SparkSubmit.prepareSubmitEnvironment(SparkSubmit.scala:334)
at org.apache.spark.deploy.SparkSubmit.submit(SparkSubmit.scala:143)
at org.apache.spark.deploy.SparkSubmit.doSubmit(SparkSubmit.scala:86)
at org.apache.spark.deploy.SparkSubmit$$anon$2.doSubmit(SparkSubmit.scala:924)
at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:933)
at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)
Caused by: java.net.UnknownHostException: c0d66fa49434: c0d66fa49434: Name does not resolve
at java.net.InetAddress.getLocalHost(InetAddress.java:1506)
at org.apache.spark.util.Utils$.findLocalInetAddress(Utils.scala:946)
at org.apache.spark.util.Utils$.org$apache$spark$util$Utils$$localIpAddress$lzycompute(Utils.scala:939)
at org.apache.spark.util.Utils$.org$apache$spark$util$Utils$$localIpAddress(Utils.scala:939)
at org.apache.spark.util.Utils$$anonfun$localCanonicalHostName$1.apply(Utils.scala:996)
at org.apache.spark.util.Utils$$anonfun$localCanonicalHostName$1.apply(Utils.scala:996)
at scala.Option.getOrElse(Option.scala:121)
at org.apache.spark.util.Utils$.localCanonicalHostName(Utils.scala:996)
at org.apache.spark.internal.config.package$.<init>(package.scala:296)
at org.apache.spark.internal.config.package$.<clinit>(package.scala)
... 18 more
Caused by: java.net.UnknownHostException: c0d66fa49434: Name does not resolve
at java.net.Inet4AddressImpl.lookupAllHostAddr(Native Method)
at java.net.InetAddress$2.lookupAllHostAddr(InetAddress.java:929)
at java.net.InetAddress.getAddressesFromNameService(InetAddress.java:1324)
at java.net.InetAddress.getLocalHost(InetAddress.java:1501)
... 27 more

有没有人通过类似的尝试取得任何成功？

标签： apache-sparkamazon-ecsaws-fargate