Details
Description
We noticed that the `hadoop-0.20-datanode start` does not report the status correctly at the end. The root cause has been that the HADOOP_IDENT_STRING was taken from /etc/default/hadoop-0.20 and not from hadoop-env.sh, so the PID file was wrong.
Should we source the hadoop-env.sh after the /etc/default/hadoop-0.20 in the init.d scripts?
Alex K