Hi,
I used TestDFSIO to test the HDFS’s performance, but I found that I
can't run concurrent map task when specifying -nrFiles larger than 1. The
command that I used is like this "hadoop jar
share/hadoop/mapreduce/hadoop-mapreduce-client-jobclient-3.0.0-tests.jar
TestDFSIO -write -nrFiles 4 -fileSize 24GB”. The command output map result’s
one by one. I also found that the files in the benchmark path of
HDFS(/benchmarks/TestDFSIO/io_data) is created one by one, too. My server of
benchmark client has 24 cores, and I want to run multiple map tasks
concurrently, how can I achieve this?
Thanks all of you!
Arvin
