在集群模式下使用spark submit时，我为file提供了--files参数当我在并行窗口中运行list命令时，我可以看到这些文件

$ hdfs dfs -ls /user/name/.sparkStaging/application_1598445422718_5631/
-rw-r--r--   3 user hadoop       1234 2020-11-18 06:44 /user/name/.sparkStaging/application_1598445422718_5631/survey.properties

我是这样读的：

val props = Source.fromFile(System.getenv("SPARK_YARN_STAGING_DIR")+"/survey.properties")

在日志中我可以看到：

survey.properties -> resource { scheme: "hdfs" host: "myhost" port: -1 file: "/user/name/.sparkStaging/application_1598445422718_5631/survey.properties" } size: 1234timestamp: 1605680821911 type: FILE visibility: PRIVATE

但仍然得到以下错误：

client token: Token { kind: YARN_CLIENT_TOKEN, service:  }
         diagnostics: User class threw exception: java.io.FileNotFoundException: hdfs:/myhost/user/name/.sparkStaging/application_1598445422718_5631/survey.properties (No such file or directory)