How does user user jar conflict resolved in spark interpreter?

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

How does user user jar conflict resolved in spark interpreter?

Serega Sheypak
Hi zeppelin users!
I have the question about dependencies users are using while running notebooks using spark interpreter.

Imagine I have configured spark intepreter.

Two users write their spark notebooks. 
the first user does 
z.load("com:best-it-company:0.1")

the second one user adds to his notebook:

z
.load("com:best-it-company:0.2
")

Then they start to execute two notebooks concurrently.
What will happen to dependencies?
They have same classes... will spark isolate 0.1 version from 0.2 version somehow?
Reply | Threaded
Open this post in threaded view
|

Re: How does user user jar conflict resolved in spark interpreter?

Jeff Zhang

You can try mode isolated per user which would create one jvm for each user.  



Serega Sheypak <[hidden email]>于2017年11月16日周四 上午5:42写道:
Hi zeppelin users!
I have the question about dependencies users are using while running notebooks using spark interpreter.

Imagine I have configured spark intepreter.

Two users write their spark notebooks. 
the first user does 
z.load("com:best-it-company:0.1")

the second one user adds to his notebook:

z
.load("com:best-it-company:0.2
")

Then they start to execute two notebooks concurrently.
What will happen to dependencies?
They have same classes... will spark isolate 0.1 version from 0.2 version somehow?