I only change the content of the jar, not the name or version of the jar 
(otherwise I’d have to re-add it as a dependency anyway).  Or do you mean 
something else by ’version’?

This dependency is a local file. Zeppelin and Spark are all running on the same 
machine. So I’m just specifying the file system path of the jar; it’s not even 
prefixed with file:///.

From: Jhon Anderson Cardenas Diaz [mailto:jhonderson2...@gmail.com]
Sent: 22 February 2018 12:18
To: users@zeppelin.apache.org
Subject: EXT: Re: Jar dependencies are not reloaded when Spark interpreter is 

When you say you change the dependency, is only about the content? Or content 
and version. I think the dependency should be reloaded only if its version 

I do not think it's optimal to re-download the dependencies every time the 
interpreter reboots.

El 22 feb. 2018 05:22, "Partridge, Lucas (GE Aviation)" 
<lucas.partri...@ge.com<mailto:lucas.partri...@ge.com>> escribió:
I’m using Zeppelin 0.7.3 against a local standalone Spark ‘cluster’. I’ve added 
a Scala jar dependency to my Spark interpreter using Zeppelin’s UI. I thought 
if I changed my Scala code and updated the jar (using sbt outside of Zeppelin) 
then all I’d have to do is restart the interpreter for the new code to be 
picked up in Zeppelin in a regular scala paragraph.  However restarting the 
interpreter appears to have no effect – the new code is not detected. Is that 
expected behaviour or a bug?

The workaround I’m using at the moment is to edit the spark interpreter, remove 
the jar, re-add it, save the changes and then restart the interpreter. Clumsy 
but that’s better than restarting Zeppelin altogether.

Also, if anyone knows of a better way to reload code without restarting the 
interpreter then I’m open to suggestions:). Having to re-run lots of paragraphs 
after a restart is pretty tedious.

Thanks, Lucas.

Reply via email to