On Sun, Jun 24, 2012 at 11:22 PM, Jean-Marc Spaggiari <jean-m...@spaggiari.org> wrote: > Hi, > > In HBASE-1512 (https://issues.apache.org/jira/browse/HBASE-1512) there > is the implementation of co-processor for count and others. > > Is there anywhere an example of the way to use them? Because the shell > count is very slow with there is to many rows. >
You've enabled bulk reading of rows, hbase.client.scanner.caching [1], as per http://hbase.apache.org/book.html#rowcounter? Pass a configuration on the command-line with -Dhbase.client.scanner.caching=1000 or if in shell, its an option to pass the count command. If it still too slow, run your mapreduce as a mapreduce job. The hbase-X.X.X.jar is also a mapreduce job jar with a few jobs bundled including rowcounter. See here for how to run it: http://hbase.apache.org/book.html#driver St.Ack 1. http://hbase.apache.org/book.html#hbase.client.scanner.caching