sh build-image.sh
docker run -i -t samklr/pyspark-notebook pyspark
docker run -i -t samklr/pyspark-notebook spark-shell
boot2docker ip
docker run -it -p 8888:8888 samklr/pyspark-notebook ipython notebook --ipython-dir=/.ipython --profile=pyspark
docker run -it -p 8888:8888 -v $(PWD):/mnt samklr/pyspark-notebook
ipython notebook --ipython-dir=/.ipython --profile=pyspark ipython kernel --ipython-dir=/.ipython --profile=pyspark
######################
sh build-image.sh docker run -it --publish-all -v $(PWD):/mnt --rm --name ipython-kernel samklr/pyspark-notebook ipython kernel --ipython-dir=/mnt/ipython --profile=pyspark
docker run -it --publish-all -v $(PWD):/mnt --link=ipython-kernel:ipython-kernel samklr/pyspark-notebook ipython console --ipython-dir=/mnt/ipython --profile=pyspark --existing kernel-19.json
-> Note, I tried changing the 'ip' in the kernel-7.json to {boot2docker ip}, but in both cases get a "ERROR: Kernel did not respond"
-
So kernel server publishes all docker run --publish-all
-
client server can then read those
-
TODO need a script which crafts a Dockerfile
-
TODO want to have json witten to volumn
-
TODO Want to look at json
-
TODO want to connect to kernel using another docker machine?
** Run linux commands: apt-get install screen ** run ipython notebook: screen -S ipynotebook ipython notebook --ipython-dir=/.ipython --profile=pyspark ** run ipython to attach kernel *** TODO How to have a kernel generated by notebook?
- TODO look to see if there is a kernel .json in the VM ** maybe I can connect if the FS is shared and ports are shared
- and then just connect to this externally
- then maybe connect via
- Ipython kernel stuff: ** first start up a kernel server: ipython kernel ** Then connect: ipython console --existing kernel-47897.json
docker run -it -p 8888:8888 -v $(PWD):/mnt samklr/pyspark-notebook ipython notebook --ipython-dir=/.ipython --profile=pyspark
-
TODO maybe just get pyspark running (in local) ** and connect to this
-
DONE Want to use ipython-notebook, ** TODO attach to kernels *** TODO could use external ipython to attach to kernel ** TODO attache to pyspark kernel
-
TODO want to upgrade spark to 1.2: https://spark.apache.org/downloads.html
#######
-
The kernel seems to crash when I try to create a new notebook
-
TODO is there some directory which is attempting to be written to?
-
TODO Can I just have a simple docker
docker run /bin/bash
============================================
######################
Datascience-dockerized
======================
Remote Hackathon
Data science Environment with IPython Notebooks, Spark Cluster via docker
Spins up a Container with Spark installed, and IPython Notebook server.
Clone
Get into pyspark-docker
Build your docker image : ** sh pyspark-docker/build-images **
Run the container with the notebook server :
**docker run -d -p 8888:8888 samklr/pyspark-notebook ipython notebook --profile=pyspark**
If you just want to play with spark from the command line :
** sudo docker run -i -t samklr/pyspark-notebook pyspark **
** sudo docker run -i -t samklr/pyspark-notebook spark-shell **
Head to your browser : * http://[your_IP_address or localhost]:8888 *
Next : Full web app with embedded notebooks and control of environments.
*Support for Scala via iscala and Andy's spark notebook*
*Full support for multiple clusters via Kubernetes (work in progress)*
*Deploy Script to Mesos/Marathon*
Sam Bessalah
@samklr