[
https://issues.apache.org/jira/browse/YARN-1721?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13902253#comment-13902253
]
Karthik Kambatla commented on YARN-1721:
----------------------------------------
The fix itself looks good to me. Good catch.
Nits:
# The following comment is not very clear - had to read a couple of times to
understand. How about simplifying it - "Synchronize on attempt to serialize
with FairScheduler#allocate"
{code}
+ // FairScheduler#allocate does not take out the global lock on the
scheduler,
+ // so synchronizing on the app here avoids moving it while its requests are
+ // being updated.
{code}
# synchronized -> synchronization
{code}
+ * Helper for moveApplication, which has appropriate synchronized, so all
{code}
> When moving app between queues in Fair Scheduler, grab lock on FSSchedulerApp
> -----------------------------------------------------------------------------
>
> Key: YARN-1721
> URL: https://issues.apache.org/jira/browse/YARN-1721
> Project: Hadoop YARN
> Issue Type: Bug
> Components: scheduler
> Reporter: Sandy Ryza
> Assignee: Sandy Ryza
> Attachments: YARN-1721.patch
>
>
> FairScheduler.moveApplication should grab lock on FSSchedulerApp, so that
> allocate() can't be modifying it at the same time.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
