[
https://issues.apache.org/jira/browse/SPARK-11509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14994616#comment-14994616
]
Andrew Davidson commented on SPARK-11509:
-----------------------------------------
okay after a couple of days hacking it looks like my test program might be
working. Here is my recipe. (I hope this helps others)
My test program is now
In [1]:
from pyspark import SparkContext
textFile = sc.textFile("file:///home/ec2-user/dataScience/readme.md")
In [2]:
print("hello world”)
hello world
In [3]:
textFile.take(3)
Out[3]:
[' hello world', '']
Installation instructions
Ssh to cluster master
Sudo su
install python3.4 on all machines
```
yum install -y python34
bash-4.2# which python3
/usr/bin/python3
pssh -h /root/spark-ec2/slaves yum install -y python34
```
4. Install pip on all machines
```
yum list available |grep pip
yum install -y python34-pip
find /usr/bin -name "*pip*" -print
/usr/bin/pip-3.4
pssh -h /root/spark-ec2/slaves yum install -y python34-pip
```
5. install python on master
```
/usr/bin/pip-3.4 install ipython
pssh -h /root/spark-ec2/slaves /usr/bin/pip-3.4 install python
```
6. Install python develop stuff and jupiter on master
```
yum install -y python34-devel
/usr/bin/pip-3.4 install jupyter
```
7. Set up update spark-env.sh on all machine so by default we use python3.4
```
cd /root/spark/conf
printf "\n# Set Spark Python version\nexport PYSPARK_PYTHON=python3.4\n" >>
/root/spark/conf/spark-env.sh
for i in `cat slaves` ; do scp spark-env.sh
root@$i:/root/spark/conf/spark-env.sh; done
```
8. Restart cluster
```
/root/spark/sbin/stop-all.sh
/root/spark/sbin/start-all.sh
```
Running ipython notebook
set up an ssh tunnel on your local machine
ssh -i $KEY_FILE -N -f -L localhost:8888:localhost:7000 ec2-user@$SPARK_MASTER
2. Log on to cluster master and start ipython notebook server
```
export PYSPARK_PYTHON=python3.4
export PYSPARK_DRIVER_PYTHON=python3.4
export IPYTHON_OPTS="notebook --no-browser --port=7000"
$SPARK_ROOT/bin/pyspark --master local[2]
```
3. On your local machine open http://localhost:8888
> ipython notebooks do not work on clusters created using
> spark-1.5.1-bin-hadoop2.6/ec2/spark-ec2 script
> ------------------------------------------------------------------------------------------------------
>
> Key: SPARK-11509
> URL: https://issues.apache.org/jira/browse/SPARK-11509
> Project: Spark
> Issue Type: Bug
> Components: Documentation, EC2, PySpark
> Affects Versions: 1.5.1
> Environment: AWS cluster
> [ec2-user@ip-172-31-29-60 ~]$ uname -a
> Linux ip-172-31-29-60.us-west-1.compute.internal 3.4.37-40.44.amzn1.x86_64 #1
> SMP Thu Mar 21 01:17:08 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
> Reporter: Andrew Davidson
>
> I recently downloaded spark-1.5.1-bin-hadoop2.6 to my local mac.
> I used spark-1.5.1-bin-hadoop2.6/ec2/spark-ec2 to create an aws cluster. I am
> able to run the java SparkPi example on the cluster how ever I am not able to
> run ipython notebooks on the cluster. (I connect using ssh tunnel)
> According to the 1.5.1 getting started doc
> http://spark.apache.org/docs/latest/programming-guide.html#using-the-shell
> The following should work
> PYSPARK_DRIVER_PYTHON=ipython PYSPARK_DRIVER_PYTHON_OPTS="notebook
> --no-browser --port=7000" /root/spark/bin/pyspark
> I am able to connect to the notebook server and start a notebook how ever
> bug 1) the default sparkContext does not exist
> from pyspark import SparkContext
> textFile = sc.textFile("file:///home/ec2-user/dataScience/readme.md")
> textFile.take(3
> ---------------------------------------------------------------------------
> NameError Traceback (most recent call last)
> <ipython-input-1-127b6a58d5cc> in <module>()
> 1 from pyspark import SparkContext
> ----> 2 textFile = sc.textFile("file:///home/ec2-user/dataScience/readme.md")
> 3 textFile.take(3)
> NameError: name 'sc' is not defined
> bug 2)
> If I create a SparkContext I get the following python versions miss match
> error
> sc = SparkContext("local", "Simple App")
> textFile = sc.textFile("file:///home/ec2-user/dataScience/readme.md")
> textFile.take(3)
> File "/root/spark/python/lib/pyspark.zip/pyspark/worker.py", line 64, in main
> ("%d.%d" % sys.version_info[:2], version))
> Exception: Python in worker has different version 2.7 than that in driver
> 2.6, PySpark cannot run with different minor versions
> I am able to run ipython notebooks on my local Mac as follows. (by default
> you would get an error that the driver and works are using different version
> of python)
> $ cat ~/bin/pySparkNotebook.sh
> #!/bin/sh
> set -x # turn debugging on
> #set +x # turn debugging off
> export PYSPARK_PYTHON=python3
> export PYSPARK_DRIVER_PYTHON=python3
> IPYTHON_OPTS=notebook $SPARK_ROOT/bin/pyspark $*$
> I have spent a lot of time trying to debug the pyspark script however I can
> not figure out what the problem is
> Please let me know if there is something I can do to help
> Andy
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
