我使用 bdutil 在 Hadoop (2.6) 集群上部署了带有 yarn-client 的 Spark (1.3.1)，默认情况下，实例是使用临时外部 ips 创建的，到目前为止 spark 工作正常。出于一些安全考虑，并假设集群只能在内部访问，我从实例中删除了外部 ip；之后，spark-shell 甚至无法运行，并且似乎无法与 Yarn/Hadoop 通信，并且无限期地卡住了。只有在我添加了外部 ips 之后，spark-shell 才开始正常工作。
我的问题是，在纱线上运行火花是否需要节点的外部 ip，为什么？如果是，是否会担心安全等问题？谢谢！
简答
您需要外部 IP 地址来访问 GCS，默认 bdutil 设置将 GCS 设置为默认 Hadoop 文件系统，包括控制文件。使用./bdutil -F hdfs ... deployHDFS 作为默认值。
除非您在 GCE 网络配置中的防火墙规则中添加了太多许可规则，否则使用外部 IP 地址时安全性不应成为问题。
编辑：目前似乎存在一个错误，spark.eventLog.dir即使 default_fs 是 hdfs，我们也设置为 GCS 路径。我提交了https://github.com/GoogleCloudPlatform/bdutil/issues/35来跟踪它。同时，只需在您的主服务器上手动编辑/home/hadoop/spark-install/conf/spark-defaults.conf（您可能需要对其sudo -u hadoop vim.tiny /home/hadoop/spark-install/conf/spark-defaults.conf拥有编辑权限）以设置spark.eventLog.dir为hdfs:///spark-eventlog-base或 HDFS 中的其他内容，然后运行hadoop fs -mkdir -p hdfs:///spark-eventlog-base以使其正常工作。
长答案
默认情况下，bdutil 还将 Google Cloud Storage 配置为“默认 Hadoop 文件系统”，这意味着 Spark 和 YARN 使用的控制文件需要访问 Google Cloud Storage。此外，需要外部 IP 才能访问 Google Cloud Storage。
在手动配置网络内 SSH 后，我确实设法部分重现了您的案例；在启动过程中，我实际上看到以下内容：
15/06/26 17:23:05 INFO yarn.Client: Preparing resources for our AM container
15/06/26 17:23:05 INFO gcs.GoogleHadoopFileSystemBase: GHFS version: 1.4.0-hadoop2
15/06/26 17:23:26 WARN http.HttpTransport: exception thrown while executing request
java.net.SocketTimeoutException: connect timed out
at java.net.PlainSocketImpl.socketConnect(Native Method)
at java.net.AbstractPlainSocketImpl.doConnect(AbstractPlainSocketImpl.java:339)
at java.net.AbstractPlainSocketImpl.connectToAddress(AbstractPlainSocketImpl.java:200)
at java.net.AbstractPlainSocketImpl.connect(AbstractPlainSocketImpl.java:182)
at java.net.SocksSocketImpl.connect(SocksSocketImpl.java:392)
at java.net.Socket.connect(Socket.java:579)
at sun.security.ssl.SSLSocketImpl.connect(SSLSocketImpl.java:625)
at sun.net.NetworkClient.doConnect(NetworkClient.java:175)
at sun.net.www.http.HttpClient.openServer(HttpClient.java:432)
at sun.net.www.http.HttpClient.openServer(HttpClient.java:527)
at sun.net.www.protocol.https.HttpsClient.<init>(HttpsClient.java:275)
at sun.net.www.protocol.https.HttpsClient.New(HttpsClient.java:371)
at sun.net.www.protocol.https.AbstractDelegateHttpsURLConnection.getNewHttpClient(AbstractDelegateHttpsURLConnection.java:191)
at sun.net.www.protocol.http.HttpURLConnection.plainConnect(HttpURLConnection.java:933)
at sun.net.www.protocol.https.AbstractDelegateHttpsURLConnection.connect(AbstractDelegateHttpsURLConnection.java:177)
at sun.net.www.protocol.https.HttpsURLConnectionImpl.connect(HttpsURLConnectionImpl.java:153)
at com.google.api.client.http.javanet.NetHttpRequest.execute(NetHttpRequest.java:93)
at com.google.api.client.http.HttpRequest.execute(HttpRequest.java:965)
at com.google.api.client.googleapis.services.AbstractGoogleClientRequest.executeUnparsed(AbstractGoogleClientRequest.java:410)
at com.google.api.client.googleapis.services.AbstractGoogleClientRequest.executeUnparsed(AbstractGoogleClientRequest.java:343)
at com.google.api.client.googleapis.services.AbstractGoogleClientRequest.execute(AbstractGoogleClientRequest.java:460)
at com.google.cloud.hadoop.gcsio.GoogleCloudStorageImpl.getBucket(GoogleCloudStorageImpl.java:1557)
at com.google.cloud.hadoop.gcsio.GoogleCloudStorageImpl.getItemInfo(GoogleCloudStorageImpl.java:1512)
at com.google.cloud.hadoop.gcsio.CacheSupplementedGoogleCloudStorage.getItemInfo(CacheSupplementedGoogleCloudStorage.java:516)
at com.google.cloud.hadoop.gcsio.GoogleCloudStorageFileSystem.getFileInfo(GoogleCloudStorageFileSystem.java:1016)
at com.google.cloud.hadoop.gcsio.GoogleCloudStorageFileSystem.exists(GoogleCloudStorageFileSystem.java:382)
at com.google.cloud.hadoop.fs.gcs.GoogleHadoopFileSystemBase.configureBuckets(GoogleHadoopFileSystemBase.java:1639)
at com.google.cloud.hadoop.fs.gcs.GoogleHadoopFileSystem.configureBuckets(GoogleHadoopFileSystem.java:71)
at com.google.cloud.hadoop.fs.gcs.GoogleHadoopFileSystemBase.configure(GoogleHadoopFileSystemBase.java:1587)
at com.google.cloud.hadoop.fs.gcs.GoogleHadoopFileSystemBase.initialize(GoogleHadoopFileSystemBase.java:776)
at com.google.cloud.hadoop.fs.gcs.GoogleHadoopFileSystemBase.initialize(GoogleHadoopFileSystemBase.java:739)
at org.apache.hadoop.fs.FileSystem.createFileSystem(FileSystem.java:2596)
at org.apache.hadoop.fs.FileSystem.access$200(FileSystem.java:91)
at org.apache.hadoop.fs.FileSystem$Cache.getInternal(FileSystem.java:2630)
at org.apache.hadoop.fs.FileSystem$Cache.get(FileSystem.java:2612)
at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:370)
at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:169)
at org.apache.spark.deploy.yarn.Client.prepareLocalResources(Client.scala:216)
at org.apache.spark.deploy.yarn.Client.createContainerLaunchContext(Client.scala:384)
at org.apache.spark.deploy.yarn.Client.submitApplication(Client.scala:102)
at org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend.start(YarnClientSchedulerBackend.scala:58)
at org.apache.spark.scheduler.TaskSchedulerImpl.start(TaskSchedulerImpl.scala:141)
at org.apache.spark.SparkContext.<init>(SparkContext.scala:381)
at org.apache.spark.repl.SparkILoop.createSparkContext(SparkILoop.scala:1016)
at $line3.$read$$iwC$$iwC.<init>(<console>:9)
at $line3.$read$$iwC.<init>(<console>:18)
at $line3.$read.<init>(<console>:20)
at $line3.$read$.<init>(<console>:24)
at $line3.$read$.<clinit>(<console>)
at $line3.$eval$.<init>(<console>:7)
at $line3.$eval$.<clinit>(<console>)
at $line3.$eval.$print(<console>)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:606)
at org.apache.spark.repl.SparkIMain$ReadEvalPrint.call(SparkIMain.scala:1065)
at org.apache.spark.repl.SparkIMain$Request.loadAndRun(SparkIMain.scala:1338)
at org.apache.spark.repl.SparkIMain.loadAndRunReq$1(SparkIMain.scala:840)
at org.apache.spark.repl.SparkIMain.interpret(SparkIMain.scala:871)
at org.apache.spark.repl.SparkIMain.interpret(SparkIMain.scala:819)
at org.apache.spark.repl.SparkILoop.reallyInterpret$1(SparkILoop.scala:856)
at org.apache.spark.repl.SparkILoop.interpretStartingWith(SparkILoop.scala:901)
at org.apache.spark.repl.SparkILoop.command(SparkILoop.scala:813)
at org.apache.spark.repl.SparkILoopInit$$anonfun$initializeSpark$1.apply(SparkILoopInit.scala:123)
at org.apache.spark.repl.SparkILoopInit$$anonfun$initializeSpark$1.apply(SparkILoopInit.scala:122)
at org.apache.spark.repl.SparkIMain.beQuietDuring(SparkIMain.scala:324)
at org.apache.spark.repl.SparkILoopInit$class.initializeSpark(SparkILoopInit.scala:122)
at org.apache.spark.repl.SparkILoop.initializeSpark(SparkILoop.scala:64)
at org.apache.spark.repl.SparkILoop$$anonfun$org$apache$spark$repl$SparkILoop$$process$1$$anonfun$apply$mcZ$sp$5.apply$mcV$sp(SparkILoop.scala:973)
at org.apache.spark.repl.SparkILoopInit$class.runThunks(SparkILoopInit.scala:157)
at org.apache.spark.repl.SparkILoop.runThunks(SparkILoop.scala:64)
at org.apache.spark.repl.SparkILoopInit$class.postInitialization(SparkILoopInit.scala:106)
at org.apache.spark.repl.SparkILoop.postInitialization(SparkILoop.scala:64)
at org.apache.spark.repl.SparkILoop$$anonfun$org$apache$spark$repl$SparkILoop$$process$1.apply$mcZ$sp(SparkILoop.scala:990)
at org.apache.spark.repl.SparkILoop$$anonfun$org$apache$spark$repl$SparkILoop$$process$1.apply(SparkILoop.scala:944)
at org.apache.spark.repl.SparkILoop$$anonfun$org$apache$spark$repl$SparkILoop$$process$1.apply(SparkILoop.scala:944)
at scala.tools.nsc.util.ScalaClassLoader$.savingContextLoader(ScalaClassLoader.scala:135)
at org.apache.spark.repl.SparkILoop.org$apache$spark$repl$SparkILoop$$process(SparkILoop.scala:944)
at org.apache.spark.repl.SparkILoop.process(SparkILoop.scala:1058)
at org.apache.spark.repl.Main$.main(Main.scala:31)
at org.apache.spark.repl.Main.main(Main.scala)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:606)
at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:569)
at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:166)
at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:189)
at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:110)
at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)
正如预期的那样，只需调用org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend.start它就尝试联系 Google Cloud Storage，但由于没有外部 IP 没有 GCS 访问而失败。
为了解决这个问题，您可以-F hdfs在创建集群时简单地使用 HDFS 作为默认文件系统；在这种情况下，即使没有外部 IP 地址，一切都应该在集群内工作。gs://bucket/object在这种模式下，只要通过将完整路径指定为 Hadoop 参数来分配外部 IP 地址，您甚至仍然可以继续使用 GCS 。但是请注意，在这种情况下，只要您删除了外部 IP 地址，您将无法使用 GCS，除非您还配置了代理服务器并通过您的代理收集所有数据；GCS 配置是fs.gs.proxy.address.
一般来说，没有必要仅仅因为拥有外部 IP 地址而担心安全性，除非您在 Google Compute Engine 的“默认”网络防火墙规则中打开了新的许可规则。