Thanks for the feedback.
> HBASE_VERSION=`echo $HBASE_HOME | cut -d '-' -f 2`
This works, so that's ok with me. I committed your suggestion. The Hadoop EC2
scripts, at least the old ones, do something similar.
> How can I re-start my Hadoop/HBase cluster through the ec2 script or the
> other easy way?
The scripts can only do a full restart:
./bin/hbase-ec2 terminate-cluster <name>
./bin/hbase-ec2 launch-cluster <name> <slaves> <zoos>
You can also log on to the master and use the HBase control scripts:
./bin/hbase-ec2 login <name>
$ cd /usr/local/hbase*
$ ./bin/stop-hbase.sh
$ ./bin/start-hbase.sh
but before that would work you must manually update
/usr/local/hbase*/conf/regionservers with all of the internal DNS names of your
cluster slaves.
- Andy
________________________________
From: stchu <[email protected]>
To: [email protected]
Sent: Fri, December 11, 2009 2:23:55 AM
Subject: Question about ec2 script
Hi,
I found one little mistake in In the hbase-ec2-init-remote.sh:
# Update classpath to include HBase jars and config
cat >> $HADOOP_HOME/conf/hadoop-env.sh <<EOF
HADOOP_CLASSPATH="$HBASE_HOME/hbase-${HBASE_VERSION}.jar:$HBASE_HOME/lib/AgileJSON-2009-03-30.jar:$HBASE_HOME/lib/json.jar:$HBASE_HOME/lib/zookeeper-3.2.1.jar:$HBASE_HOME/conf"
EOF
${HBASE_VERSION} seem's undefine here.
I simply add a line: HBASE_VERSION=`echo $HBASE_HOME | cut -d '-' -f 2`
before that block, and that's work correctly (although this is not a good
way :P).
Besides, I have a small question:
How can I re-start my Hadoop/HBase cluster through the ec2 script or the
other easy way?
Thanks a lot!
stchu
