Github user colorant commented on the pull request:
https://github.com/apache/spark/pull/1103#issuecomment-46271800
Though with this patch, it is still possible that in the even rare case,
the info is removed, and then add back again before this synchronized block is
entered. then the read data operation should be ok, but we are synchronized on
different info instance. This could hardly been resolved. I think unless you
enlarge the synchronized scope ( which might not be efficient ) or the
blockInfo.get/put operation could employ read/write lock base on every single
entry.
---
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.
---
