Github user liancheng commented on the pull request:
https://github.com/apache/spark/pull/3149#issuecomment-62107397
@OopsOutOfMemory The `map(_.copy())` makes sense, because `HiveTableScan`
uses single a mutable row object to traverse the underlying table for
optimization purposes (reducing object number and GC pressure). If you collect
the RDD without copying, all row object reference in a single RDD partition
point to the same mutable row object. However, this situation should have been
dealt properly before. I'm tracking how and when this bug was introduced.
And you should be able to reproduce this issue with the current master
branch. Note that only SBT build can be used until #3105 and #3103 are merged.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
