[
https://issues.apache.org/jira/browse/HADOOP-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12715527#action_12715527
]
Steve Loughran commented on HADOOP-5921:
----------------------------------------
I have done a refactoring of the JT constructor in the 3628 service lifecycle
stuff. So I am in a position to interrupt things there, even though nobody else
is.
Looking at this code, I don't think it will make things any harder to
interrupt, and it does simplify some of those boostrap race conditions. I
effectively have to do that same blocking until >1 worker is live before
pushing files or work out; having this built in would simplify stuff.
There is (already) a bit of a problem with trying to kill a JT before it is
live; this doesn't show up when you start Hadoop via the scripts, but I've seen
it. I don't think this makes anything worse.
> JobTracker does not come up because of NotReplicatedYetException
> ----------------------------------------------------------------
>
> Key: HADOOP-5921
> URL: https://issues.apache.org/jira/browse/HADOOP-5921
> Project: Hadoop Core
> Issue Type: Bug
> Components: mapred
> Affects Versions: 0.20.0
> Reporter: Amareshwari Sriramadasu
> Fix For: 0.20.1
>
> Attachments: HADOOP-5921-v1.0.patch
>
>
> Sometimes (On a big cluster) Jobtracker does not come up, because Info file
> could not be replicated on dfs.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
