[ http://issues.apache.org/jira/browse/HADOOP-285?page=all ]
Hairong Kuang updated HADOOP-285:
---------------------------------
Attachment: datanode.patch
This patch starts the data receiver thread in "run" instead of "offerservice".
So it will not be restarted after a connect is lost.
> Data nodes cannot re-join the cluster once connection is lost
> -------------------------------------------------------------
>
> Key: HADOOP-285
> URL: http://issues.apache.org/jira/browse/HADOOP-285
> Project: Hadoop
> Type: Bug
> Components: dfs
> Versions: 0.3.0
> Reporter: Konstantin Shvachko
> Assignee: Hairong Kuang
> Attachments: datanode.patch
>
> A data node looses connection to a name node and then tries to offerService()
> again.
> HADOOP-270 changes force it to start dataXceiveServer, which is already
> started and in this case
> throws IllegalThreadStateException, which goes on in a loop, and never
> reaches the heartbeat section.
> So the data node never re-joins the cluster, while from the out side it looks
> it's still running.
> This is another reason why we see missing data, and don't see failed data
> nodes.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira
