diff options
author | Davies Liu <davies@databricks.com> | 2014-11-14 20:13:46 -0800 |
---|---|---|
committer | Andrew Or <andrew@databricks.com> | 2014-11-14 20:14:33 -0800 |
commit | 7fe08b43c78bf9e8515f671e72aa03a83ea782f8 (patch) | |
tree | 9a8fbee009b7f6fc52f11092d90bdb5ace361df0 | |
parent | 303a4e4d23e5cd93b541480cf88d5badb9cf9622 (diff) | |
download | spark-7fe08b43c78bf9e8515f671e72aa03a83ea782f8.tar.gz spark-7fe08b43c78bf9e8515f671e72aa03a83ea782f8.tar.bz2 spark-7fe08b43c78bf9e8515f671e72aa03a83ea782f8.zip |
[SPARK-4415] [PySpark] JVM should exit after Python exit
When JVM is started in a Python process, it should exit once the stdin is closed.
test: add spark.driver.memory in conf/spark-defaults.conf
```
daviesdm:~/work/spark$ cat conf/spark-defaults.conf
spark.driver.memory 8g
daviesdm:~/work/spark$ bin/pyspark
>>> quit
daviesdm:~/work/spark$ jps
4931 Jps
286
daviesdm:~/work/spark$ python wc.py
943738
0.719928026199
daviesdm:~/work/spark$ jps
286
4990 Jps
```
Author: Davies Liu <davies@databricks.com>
Closes #3274 from davies/exit and squashes the following commits:
df0e524 [Davies Liu] address comments
ce8599c [Davies Liu] address comments
050651f [Davies Liu] JVM should exit after Python exit
-rwxr-xr-x | bin/pyspark | 2 | ||||
-rw-r--r-- | bin/pyspark2.cmd | 1 | ||||
-rw-r--r-- | core/src/main/scala/org/apache/spark/deploy/SparkSubmitDriverBootstrapper.scala | 11 | ||||
-rw-r--r-- | python/pyspark/java_gateway.py | 4 |
4 files changed, 9 insertions, 9 deletions
diff --git a/bin/pyspark b/bin/pyspark index 1d8c94d43d..0b4f695dd0 100755 --- a/bin/pyspark +++ b/bin/pyspark @@ -132,7 +132,5 @@ if [[ "$1" =~ \.py$ ]]; then gatherSparkSubmitOpts "$@" exec "$FWDIR"/bin/spark-submit "${SUBMISSION_OPTS[@]}" "$primary" "${APPLICATION_OPTS[@]}" else - # PySpark shell requires special handling downstream - export PYSPARK_SHELL=1 exec "$PYSPARK_DRIVER_PYTHON" $PYSPARK_DRIVER_PYTHON_OPTS fi diff --git a/bin/pyspark2.cmd b/bin/pyspark2.cmd index 59415e9bde..a542ec80b4 100644 --- a/bin/pyspark2.cmd +++ b/bin/pyspark2.cmd @@ -59,7 +59,6 @@ for /f %%i in ('echo %1^| findstr /R "\.py"') do ( ) if [%PYTHON_FILE%] == [] ( - set PYSPARK_SHELL=1 if [%IPYTHON%] == [1] ( ipython %IPYTHON_OPTS% ) else ( diff --git a/core/src/main/scala/org/apache/spark/deploy/SparkSubmitDriverBootstrapper.scala b/core/src/main/scala/org/apache/spark/deploy/SparkSubmitDriverBootstrapper.scala index 7ffff29122..aa3743ca7d 100644 --- a/core/src/main/scala/org/apache/spark/deploy/SparkSubmitDriverBootstrapper.scala +++ b/core/src/main/scala/org/apache/spark/deploy/SparkSubmitDriverBootstrapper.scala @@ -149,14 +149,15 @@ private[spark] object SparkSubmitDriverBootstrapper { // subprocess there already reads directly from our stdin, so we should avoid spawning a // thread that contends with the subprocess in reading from System.in. val isWindows = Utils.isWindows - val isPySparkShell = sys.env.contains("PYSPARK_SHELL") + val isSubprocess = sys.env.contains("IS_SUBPROCESS") if (!isWindows) { val stdinThread = new RedirectThread(System.in, process.getOutputStream, "redirect stdin") stdinThread.start() - // For the PySpark shell, Spark submit itself runs as a python subprocess, and so this JVM - // should terminate on broken pipe, which signals that the parent process has exited. In - // Windows, the termination logic for the PySpark shell is handled in java_gateway.py - if (isPySparkShell) { + // Spark submit (JVM) may run as a subprocess, and so this JVM should terminate on + // broken pipe, signaling that the parent process has exited. This is the case if the + // application is launched directly from python, as in the PySpark shell. In Windows, + // the termination logic is handled in java_gateway.py + if (isSubprocess) { stdinThread.join() process.destroy() } diff --git a/python/pyspark/java_gateway.py b/python/pyspark/java_gateway.py index 9c70fa5c16..a975dc19cb 100644 --- a/python/pyspark/java_gateway.py +++ b/python/pyspark/java_gateway.py @@ -45,7 +45,9 @@ def launch_gateway(): # Don't send ctrl-c / SIGINT to the Java gateway: def preexec_func(): signal.signal(signal.SIGINT, signal.SIG_IGN) - proc = Popen(command, stdout=PIPE, stdin=PIPE, preexec_fn=preexec_func) + env = dict(os.environ) + env["IS_SUBPROCESS"] = "1" # tell JVM to exit after python exits + proc = Popen(command, stdout=PIPE, stdin=PIPE, preexec_fn=preexec_func, env=env) else: # preexec_fn not supported on Windows proc = Popen(command, stdout=PIPE, stdin=PIPE) |