Github user JoshRosen commented on the pull request:
https://github.com/apache/spark/pull/1542#issuecomment-50196891
As I [commented on the
JIRA](https://issues.apache.org/jira/browse/SPARK-2638?focusedCommentId=14074710&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14074710),
unless I've missed something I'm not sure that the original code exhibits
significant concurrency issues: no time-consuming work is performed in the
synchronized blocks and the actual MapOuputTracker requests proceed in parallel
without holding any locks. The only potential issue that I see is many
concurrent duplicate requests causing a big queue of waiters on the lock that
results in many spurious wakeups as statuses are fetched; this seems unlikely
to happen in practice, though.
Do you have an actual use-case that demonstrates that the current
synchronization strategy is a performance issue?
---
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.
---
