LuciferYang commented on pull request #34881:
URL: https://github.com/apache/spark/pull/34881#issuecomment-994261211


   > May I ask you to run a maven build as well? Just to be in the safe side... 
I am not 100% sure but as you changed the exclusion in `SparkBuild.scala` but 
not in a pom.xml we might find something here.
   
   > And agree, I think you need Maven exclusions to match SBT
   
   @attilapiros @srowen  
   
   In SPARK-35134 (https://github.com/apache/spark/pull/32230), there is a 
netty version conflict problem, the phenomenon that triggers the bug is that 
even if we use `netty-4.1.51-all.jar`, there will be `copying 
deps/jars/netty-handler-4.1.50.Final.jar -> pyspark-3.2.0.dev0/deps/jars` and 
`copying deps/jars/netty-transport-native-epoll-4.1.50.Final.jar -> 
pyspark-3.2.0.dev0/deps/jars` behavior during sbt testing.
   
   So  I add 2 new `ExclusionRule` in SPARK-35134 to fix this sbt test bug.
   
   ```
   ExclusionRule("io.netty", "netty-handler"),
   ExclusionRule("io.netty", "netty-transport-native-epoll")
   ```
   
   We didn't find this problem during Maven testing, so I didn't make 
corresponding change for Maven In SPARK-35134.
   
   Then after Netty-4.1.69 ([Netty#11732 netty-all should not re-package 
jars](https://github.com/netty/netty/pull/11732)) , netty-all replaces a fat 
jar with separate modules, keeping `ExclusionRule("io.netty", "netty-handler")` 
and `ExclusionRule("io.netty", "netty-transport-native-epoll")` will result in 
the absence of `netty-transport-native-epoll-4.1.71.Final.jar` and 
`netty-handler-4.1.71.Final.jar` in the during sbt test because 
`netty-all-4.1.71.jar` no longer contains them.
   
   
   
   
   
   
   
   
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to