Merge pull request #112 from parente/fix-python2-path

Set PYSPARK_PYTHON path in python2 kernelspec

all-spark-notebook/Dockerfile

@@ -6,6 +6,9 @@ MAINTAINER Jupyter Project <jupyter@googlegroups.com>
 
 USER root
 
+# Util to help with kernel spec later
+RUN apt-get -y update && apt-get -y install jq
+
 # Spark dependencies
 ENV APACHE_SPARK_VERSION 1.5.1
 RUN apt-get -y update && \
@@ -90,12 +93,13 @@ RUN conda install --yes \
 RUN mkdir -p /opt/conda/share/jupyter/kernels/scala
 COPY kernel.json /opt/conda/share/jupyter/kernels/scala/
 
-USER root
-
-# Install Python 2 kernel spec globally to avoid permission problems when NB_UID
-# switching at runtime.
-RUN $CONDA_DIR/envs/python2/bin/python \
-    $CONDA_DIR/envs/python2/bin/ipython \
-    kernelspec install-self
-
-USER jovyan
+# Install Python 2 kernel spec into the Python 3 conda environment which
+# runs the notebook server
+RUN bash -c '. activate python2 && \
+    python -m ipykernel.kernelspec --prefix=$CONDA_DIR && \
+    . deactivate'
+# Set PYSPARK_HOME in the python2 spec
+RUN jq --arg v "$CONDA_DIR/envs/python2/bin/python" \
+        '.["env"]["PYSPARK_PYTHON"]=$v' \
+        $CONDA_DIR/share/jupyter/kernels/python2/kernel.json > /tmp/kernel.json && \
+        mv /tmp/kernel.json $CONDA_DIR/share/jupyter/kernels/python2/kernel.json

all-spark-notebook/README.md

@@ -32,7 +32,7 @@ This configuration is nice for using Spark on small, local data.
 1. Open a Python 2 or 3 notebook.
 2. Create a `SparkContext` configured for local mode.
 
-For example, the first few cells in a Python 3 notebook might read:
+For example, the first few cells in a notebook might read:
 
 ```python
 import pyspark
@@ -43,15 +43,6 @@ rdd = sc.parallelize(range(1000))
 rdd.takeSample(False, 5)
 ```
 
-In a Python 2 notebook, prefix the above with the following code to ensure the local workers use Python 2 as well.
-
-```python
-import os
-os.environ['PYSPARK_PYTHON'] = 'python2'
-
-# include pyspark cells from above here ...
-```
-
 ### In a R Notebook
 
 0. Run the container as shown above.
@@ -100,7 +91,7 @@ This configuration allows your compute cluster to scale with your data.
 
 0. Open a Python 2 or 3 notebook.
 1. Create a `SparkConf` instance in a new notebook pointing to your Mesos master node (or Zookeeper instance) and Spark binary package location.
-2. Create a `SparkContext` using this configuration. 
+2. Create a `SparkContext` using this configuration.
 
 For example, the first few cells in a Python 3 notebook might read:
 
@@ -115,7 +106,7 @@ conf = pyspark.SparkConf()
 # point to mesos master or zookeeper entry (e.g., zk://10.10.10.10:2181/mesos)
 conf.setMaster("mesos://10.10.10.10:5050")
 # point to spark binary package in HDFS or on local filesystem on all slave
-# nodes (e.g., file:///opt/spark/spark-1.5.1-bin-hadoop2.6.tgz) 
+# nodes (e.g., file:///opt/spark/spark-1.5.1-bin-hadoop2.6.tgz)
 conf.set("spark.executor.uri", "hdfs://10.10.10.10/spark/spark-1.5.1-bin-hadoop2.6.tgz")
 # set other options as desired
 conf.set("spark.executor.memory", "8g")

pyspark-notebook/Dockerfile

@@ -6,6 +6,9 @@ MAINTAINER Jupyter Project <jupyter@googlegroups.com>
 
 USER root
 
+# Util to help with kernel spec later
+RUN apt-get -y update && apt-get -y install jq
+
 # Spark dependencies
 ENV APACHE_SPARK_VERSION 1.5.1
 RUN apt-get -y update && \
@@ -52,13 +55,13 @@ RUN conda create -p $CONDA_DIR/envs/python2 python=2.7 \
     pyzmq \
     && conda clean -yt
 
-USER root
-
-# Install Python 2 kernel spec globally to avoid permission problems when NB_UID
-# switching at runtime.
-RUN $CONDA_DIR/envs/python2/bin/python \
-    $CONDA_DIR/envs/python2/bin/ipython \
-    kernelspec install-self
-
-USER jovyan
-
+# Install Python 2 kernel spec into the Python 3 conda environment which
+# runs the notebook server
+RUN bash -c '. activate python2 && \
+    python -m ipykernel.kernelspec --prefix=$CONDA_DIR && \
+    . deactivate'
+# Set PYSPARK_HOME in the python2 spec
+RUN jq --arg v "$CONDA_DIR/envs/python2/bin/python" \
+        '.["env"]["PYSPARK_PYTHON"]=$v' \
+        $CONDA_DIR/share/jupyter/kernels/python2/kernel.json > /tmp/kernel.json && \
+        mv /tmp/kernel.json $CONDA_DIR/share/jupyter/kernels/python2/kernel.json

pyspark-notebook/README.md

@@ -27,7 +27,7 @@ This configuration is nice for using Spark on small, local data.
 2. Open a Python 2 or 3 notebook.
 3. Create a `SparkContext` configured for local mode.
 
-For example, the first few cells in a Python 3 notebook might read:
+For example, the first few cells in the notebook might read:
 
 ```python
 import pyspark
@@ -38,15 +38,6 @@ rdd = sc.parallelize(range(1000))
 rdd.takeSample(False, 5)
 ```
 
-In a Python 2 notebook, prefix the above with the following code to ensure the local workers use Python 2 as well.
-
-```python
-import os
-os.environ['PYSPARK_PYTHON'] = 'python2'
-
-# include pyspark cells from above here ...
-```
-
 ## Connecting to a Spark Cluster on Mesos
 
 This configuration allows your compute cluster to scale with your data.
@@ -58,7 +49,7 @@ This configuration allows your compute cluster to scale with your data.
     * NOTE: When using `--net=host`, you must also use the flags `--pid=host -e TINI_SUBREAPER=true`. See https://github.com/jupyter/docker-stacks/issues/64 for details.
 4. Open a Python 2 or 3 notebook.
 5. Create a `SparkConf` instance in a new notebook pointing to your Mesos master node (or Zookeeper instance) and Spark binary package location.
-6. Create a `SparkContext` using this configuration. 
+6. Create a `SparkContext` using this configuration.
 
 For example, the first few cells in a Python 3 notebook might read:
 
@@ -73,7 +64,7 @@ conf = pyspark.SparkConf()
 # point to mesos master or zookeeper entry (e.g., zk://10.10.10.10:2181/mesos)
 conf.setMaster("mesos://10.10.10.10:5050")
 # point to spark binary package in HDFS or on local filesystem on all slave
-# nodes (e.g., file:///opt/spark/spark-1.5.1-bin-hadoop2.6.tgz) 
+# nodes (e.g., file:///opt/spark/spark-1.5.1-bin-hadoop2.6.tgz)
 conf.set("spark.executor.uri", "hdfs://10.122.193.209/spark/spark-1.5.1-bin-hadoop2.6.tgz")
 # set other options as desired
 conf.set("spark.executor.memory", "8g")