Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Hadoop >> mail # user >> Jobtracker still finding old task nodes


Copy link to this message
-
Re: Jobtracker still finding old task nodes
My mistake, i had assumed the trackers were not running on those machines.
But they were ...

On Sun, Aug 16, 2009 at 2:19 AM, Saptarshi Guha <[EMAIL PROTECTED]>wrote:

> Hello,
> After formatting the HDFS and removing several entries from the slaves
> file, when I start up
> $HADOOP/bin/start-all.sh (hadoop 0.20)
> I get this in jobtracker.log
> All the machines save acrux have been removed the from the slaves file.
> Both acrux and the jobtracker have the same conf files.
>
> Why does it discover the old machines? Does it automatically discover
> new machines?
>
> Thanks and Regards
> Saptarshi
>
> WARN org.apache.hadoop.mapred.JobTracker: Serious problem, cannot find
> record of 'previous' heartbeat for 'tracker_deneb.'; reinitializing
> the tasktracker
> WARN org.apache.hadoop.mapred.JobTracker: Serious problem, cannot find
> record of 'previous' heartbeat for
> 'tracker_adhara.stat.purdue.edu:localhost.localdomain/127.0.0.1:37715';
> reinitializing the tasktracker
> WARN org.apache.hadoop.mapred.JobTracker: Serious problem, cannot find
> record of 'previous' heartbeat for 'tracker_castor'; reinitializing
> the
> tasktracker
> 2009-08-16 02:15:03,146 INFO org.apache.hadoop.net.NetworkTopology:
> Adding a new node: /default-rack/deneb.
> 2009-08-16 02:15:03,161 INFO org.apache.hadoop.net.NetworkTopology:
> Adding a new node: /default-rack/adhara
> 2009-08-16 02:15:03,165 INFO org.apache.hadoop.net.NetworkTopology:
> Adding a new node: /default-rack/castor
> 2009-08-16 02:15:03,193 INFO org.apache.hadoop.net.NetworkTopology:
> Adding a new node: /default-rack/acrux.
> 2009-08-16 02:15:15,158 ERROR org.apache.hadoop.mapred.PoolManager:
> Failed to reload allocations file - will use existing allocation
>