Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Won't Fix
-
None
-
None
-
None
-
None
Description
Adding JARs to the distributed cache required a trick of FileSystem pre-initialization in the HadoopAccessor because the addFileToClasspath() method was not taking a configuration and it was creating internally a filesystem with a default configuration.
Since then the DistributedCache API as a method that takes a configuration when adding a JAR to it.
We should move to use that method as there are some edge scenarios when kerberos that this is failing.
Attachments
Issue Links
- is blocked by
-
MAPREDUCE-950 Clean up unsymmetrical Job-API w.r.t DistributedCache
- Open