Update

Author: Sephiroth-Lin

docs/submitting-applications.md

@@ -23,7 +23,7 @@ files to be distributed with your application. If you depend on multiple Python
 packaging them into a `.zip` or `.egg`.
 
 As Python can not read files from assembly jar which packaged by JDK1.7+, so packaging pyspark into a
-`.zip`(the name contains "spark-pyspark") and use `--py-files` argument of `spark-submit` to distribute it.
+`.zip`(the name contains "pyspark") and use `--py-files` argument of `spark-submit` to distribute it.
 
 # Launching Applications with spark-submit
 

yarn/src/main/scala/org/apache/spark/deploy/yarn/ExecutorRunnable.scala

@@ -78,8 +78,8 @@ class ExecutorRunnable(
     // From SPARK-1920 and SPARK-1520 we know PySpark on Yarn can not work when the assembly jar are
     // package by JDK 1.7+, so we ship PySpark archives to executors by Yarn with --py-files, and
     // add this path to PYTHONPATH.
-    for ((k, v) <- localResources if k.contains("spark-pyspark")) {
-      env("PYSPARK_ARCHIVES_PATH") = k
+    for ((resPath, res) <- localResources if resPath.contains("pyspark")) {
+      env("PYSPARK_ARCHIVES_PATH") = resPath
     }
     ctx.setEnvironment(env)