Interesting, while your network topology is changing, the node is
still running a job? 

Topology is more important for HDFS. Restarting the processes 
should re-register the node. 

-rajive

Ben Clay wrote on 01/06/12 at 08:27:55 -0800:
>   Hi-
>
>   If my network topology changes while a Hadoop job is running, is there any
>   way to re-trigger the topology script for each node?
>
>   Looking in the code, it appears JobTracker#resolveAndAddToTopology is only
>   called in JobTracker#addNewTracker, so I don’t see how this would be
>   accomplished.  However, if there are any tricks available, such as forcing
>   a slave to miss a heartbeat, I could use that.
>
>   Thanks!
>
>   -Ben
>
>

Attachment: pgpPEVZD0QDwf.pgp
Description: PGP signature

Reply via email to