apache-spark-node icon indicating copy to clipboard operation
apache-spark-node copied to clipboard

Write up instructions for using with a notebook

Open henridf opened this issue 9 years ago • 6 comments

Beaker seems like the best candidate, given its support for javascript and its polyglot nature.

My guess/hope is that this should be pretty simple...

henridf avatar Dec 17 '15 23:12 henridf

Thanks for this project.

May I suggest also looking into Apache Zeppelin? https://zeppelin.incubator.apache.org/

itayw avatar Jan 16 '16 11:01 itayw

@itayw As far as I know Zeppelin already incorporates Spark. So what exactly do you expect?

tobilg avatar Jan 16 '16 16:01 tobilg

Yep, they do have native support for Spark.

I'm still learning about this project and when, so this issue maybe out of place. However, when I saw this ticket I thought it would be great to somehow combine between Zeppelin and its capabilities and having the freedom to use JavaScript instead of Java, Scala or python. On Jan 16, 2016 6:41 PM, "Tobi" [email protected] wrote:

@itayw https://github.com/itayw As far as I know Zeppelin already incorporate Spark. So what exactly do you expect?

— Reply to this email directly or view it on GitHub https://github.com/henridf/apache-spark-node/issues/22#issuecomment-172225276 .

itayw avatar Jan 16 '16 17:01 itayw

@itayw yes Zeppelin seems like another good candidate. I don't know if it already has javascript support though? Whichever one (Zeppelin, Beaker, ...) my hope is that this shouldn't require any dev effort, just putting the pieces together...

henridf avatar Jan 16 '16 18:01 henridf

Here's a gist for building a Docker image that will allow you to run beaker with apache-node-spark.

https://gist.github.com/itayw/553effc4ee0cff5d305e

I've used beaker's image coupled with instructions taken from apache-node-spark Dockerfile. I managed to run snippets based on test/smokeTest.js without special issues.

I'm continuing to learn more about the integration and ease-of-use. Please let me know of any feedback.

itayw avatar Jan 21 '16 07:01 itayw

@itayw Thanks, looks good after a short first look. I currently update the standard Dockerfile concerning new Spark and Node.js versions. I'll open a PR for this today.

Going forward, I'd suggest to have two different Docker images, one for the standard REPL, and one for the Beaker notebook. @henridf could publish both on Docker Hub.

tobilg avatar Jan 21 '16 07:01 tobilg