+1 I use this method when performance testing on different data sets. I have several datasets to test on (varying sizes, etc). When I want to switch datasets I just shut down hbase and rename the /hbase directory...
e.g. (assuming hbase is not running) hadoop/bin/hadoop fs -mv /hbase /hbase.small hadoop/bin/hadoop fs -mv /hbase.large /hbase When I want to move my data between clusters I use: hadoop/bin/hadoop fs -copyToLocal /hbase.large /tmp/hbase.large scp -r /tmp/hbase.large u...@host:/tmp ssh u...@host hadoop/bin/hadoop fs -put /tmp/hbase.large /hbase Very handy :) On 10 February 2010 13:50, Ryan Rawson <[email protected]> wrote: > If you stop the source cluster then you can distcp the /hbase to the > other cluster. Done. A perfect copy. > > That is probably the most efficient/highest performing way. > > On Tue, Feb 9, 2010 at 6:47 PM, James Baldassari <[email protected]> wrote: > > Hi, > > > > I'm wondering if it's possible to export all data from one HBase cluster > > and import it into another. We have a lot of data that we've imported > > into our staging HBase environment, and rather than repeating the > > lengthy import process in our production environment we would prefer to > > just copy all the data directly from HBase/HDFS in staging into > > production. Is there an easy way to do this? I know Hadoop has some > > distributed copy functionality, but I don't know if this will work with > > HBase. The number of region servers and the replication factor will be > > the same in the source and destination environments, but the > > hostnames/IPs will be different. The production environment is > > completely empty right now, so we don't need to worry about overwriting > > data. > > > > I came across these links while searching for information HBase > > export/import: > > > > http://issues.apache.org/jira/browse/HBASE-897 > > http://issues.apache.org/jira/browse/HBASE-1684 > > > http://hadoop.apache.org/hbase/docs/current/api/org/apache/hadoop/hbase/mapreduce/Export.html > > > > Has anyone used these tools? Is there a better way? > > > > Thanks, > > James > > > > > > >
