Dead Node and Under Replication

[Seraph Imalia <se...@eisp.co.za>]

Hi Guys,

How do I tell hadoop to wait longer before fixing Under-Replication when 
a node dies.  Currently I have a node down which crashed, and hadoop 
detected it down and is busy fixing Under-Replication.  The problem is 
that the extra load required to fix replication is causing our 
application not to function correctly.  I want to somehow tell hadoop to 
stop fixing the Under-Replication until a non-peak time.

Is there like a timeout or a threshold setting I can set to tell hadoop 
not to fix replication immediately, but rather to wait for a period of 
time?  Anything will be better than doing it during a peak time.

Hadoop Version: 1.0.1
dfs.replication = 2
datanodes = 3

Regards,
Seraph