You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Hyukjin Kwon (Jira)" <ji...@apache.org> on 2022/06/22 03:01:00 UTC
[jira] [Commented] (SPARK-39549) How to get access to the data created in different Spark Applications
[ https://issues.apache.org/jira/browse/SPARK-39549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17557202#comment-17557202 ]
Hyukjin Kwon commented on SPARK-39549:
--------------------------------------
You should either write down into a file or a table, and read it in a different Spark application. Or, have to implement a logic to share one Spark session (e.g., zeppelin does).
> How to get access to the data created in different Spark Applications
> ---------------------------------------------------------------------
>
> Key: SPARK-39549
> URL: https://issues.apache.org/jira/browse/SPARK-39549
> Project: Spark
> Issue Type: Question
> Components: Pandas API on Spark, Project Infra, PySpark
> Affects Versions: 3.3.0
> Reporter: Chenyang Zhang
> Priority: Major
>
> I am working on a project using PySpark and I am blocked because I want to share data between different Spark applications. The situation is that we have a running java server which can handles incoming requests with a thread pool, and each thread has a corresponding python process. We want to use pandas on Spark, but have it so that any of the python processes can access the same data in spark. For example, in a python process, we created a SparkSession, read some data, modified the data using pandas on Spark api and we want to get access to that data in a different python process. The core problem is how to share data between different SparkSession or how to let different python process connect to the same SparkSession. I researched a bit but it seems impossible to share data between different python process without using external DB or connect to the same SparkSession. Generally, is this possible / what would be the recommended way to do this with the least impact on performance?
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org