After a bit of digging I found that my issue seems to be related to https://issues.apache.org/jira/browse/MAPREDUCE-2980. The task trackers get killed but the data nodes are fine.
The logs show : 2013-12-10 10:07:16,179 FATAL org.apache.hadoop.mapred.JettyBugMonitor: ************************************************************ Jetty CPU usage: 46063220198.5%. This is greater than the fatal threshold mapred.tasktracker.jetty.cpu.threshold.fatal. Aborting JVM. ************************************************************ so not really a Whirr issue as such. Julien On 10 December 2013 11:14, Julien Nioche <[email protected]>wrote: > Hi, > > I am using Whirr to launch a Hadoop 1.2.1 cluster on EC2. The cluster is > progressively losing slaves up to the point where it does not have any > left, the slave instances are still alive and running though. > > I read somewhere that the conf/slaves file is not used by Whirr so I can't > just add them back. > > Any idea of what could be wrong? > > Thanks > > Julien > > -- > > Open Source Solutions for Text Engineering > > http://digitalpebble.blogspot.com/ > http://www.digitalpebble.com > http://twitter.com/digitalpebble > -- Open Source Solutions for Text Engineering http://digitalpebble.blogspot.com/ http://www.digitalpebble.com http://twitter.com/digitalpebble
