Hi John We have encountered this problem while developing one of the features. I suggest we can raise an issue in JIRA. The problem here is there is a thread pool executor that tries to split the store files and it takes the no of store files as the no of threads needed for the executor. So for an empty region the no of store files is going to be 0 anyway.
Regards Ram > -----Original Message----- > From: saint....@gmail.com [mailto:saint....@gmail.com] On Behalf Of > Stack > Sent: Friday, September 21, 2012 10:07 AM > To: user@hbase.apache.org > Subject: Re: IllegalArgumentException when trying to split an empty > region > > On Thu, Sep 20, 2012 at 10:36 AM, John Edstrom > <jedst...@nearinfinity.com> wrote: > > Apologies, this sent before I had finished writing it :X > > > > The stack trace is below, but what we are attempting to do is load > data > > into HBase via MapReduce. When we're doing the load, we write the > HFiles > > using HFileOutputFormat and then bulk load them into HBase. At the > start of > > the MapReduce, we don't have the region splits available to us, so > all our > > data is getting written to a handful of regions. Therefore, we are > > attempting to manually split the regions before the MapReduce job so > that > > the HFiles will be evenly written across many regions. > > > > Any help or guidance would be appreciated. > > > > What version John? > > Could you use this method when you create the table: > http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HBaseAdm > in.html#createTable(org.apache.hadoop.hbase.HTableDescriptor, > byte[][]) > > St.Ack
